- title: 'AReS and MaRS Adversarial and MMD-Minimizing Regression for SDEs'
abstract: 'Stochastic differential equations are an important modeling class in many disciplines. Consequently, there exist many methods relying on various discretization and numerical integration schemes. In this paper, we propose a novel, probabilistic model for estimating the drift and diffusion given noisy observations of the underlying stochastic system. Using state-of-the-art adversarial and moment matching inference techniques, we avoid the discretization schemes of classical approaches. This leads to significant improvements in parameter accuracy and robustness given random initial guesses. On four established benchmark systems, we compare the performance of our algorithms to state-of-the-art solutions based on extended Kalman filtering and Gaussian processes.'
volume: 97
URL: https://proceedings.mlr.press/v97/abbati19a.html
PDF: http://proceedings.mlr.press/v97/abbati19a/abbati19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-abbati19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Gabriele
family: Abbati
- given: Philippe
family: Wenk
- given: Michael A.
family: Osborne
- given: Andreas
family: Krause
- given: Bernhard
family: Schölkopf
- given: Stefan
family: Bauer
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1-10
id: abbati19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1
lastpage: 10
published: 2019-05-24 00:00:00 +0000
- title: 'Dynamic Weights in Multi-Objective Deep Reinforcement Learning'
abstract: 'Many real-world decision problems are characterized by multiple conflicting objectives which must be balanced based on their relative importance. In the dynamic weights setting the relative importance changes over time and specialized algorithms that deal with such change, such as a tabular Reinforcement Learning (RL) algorithm by Natarajan and Tadepalli (2005), are required. However, this earlier work is not feasible for RL settings that necessitate the use of function approximators. We generalize across weight changes and high-dimensional inputs by proposing a multi-objective Q-network whose outputs are conditioned on the relative importance of objectives and we introduce Diverse Experience Replay (DER) to counter the inherent non-stationarity of the Dynamic Weights setting. We perform an extensive experimental evaluation and compare our methods to adapted algorithms from Deep Multi-Task/Multi-Objective Reinforcement Learning and show that our proposed network in combination with DER dominates these adapted algorithms across weight change scenarios and problem domains.'
volume: 97
URL: https://proceedings.mlr.press/v97/abels19a.html
PDF: http://proceedings.mlr.press/v97/abels19a/abels19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-abels19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Axel
family: Abels
- given: Diederik
family: Roijers
- given: Tom
family: Lenaerts
- given: Ann
family: Nowé
- given: Denis
family: Steckelmacher
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 11-20
id: abels19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 11
lastpage: 20
published: 2019-05-24 00:00:00 +0000
- title: 'MixHop: Higher-Order Graph Convolutional Architectures via Sparsified Neighborhood Mixing'
abstract: 'Existing popular methods for semi-supervised learning with Graph Neural Networks (such as the Graph Convolutional Network) provably cannot learn a general class of neighborhood mixing relationships. To address this weakness, we propose a new model, MixHop, that can learn these relationships, including difference operators, by repeatedly mixing feature representations of neighbors at various distances. MixHop requires no additional memory or computational complexity, and outperforms on challenging baselines. In addition, we propose sparsity regularization that allows us to visualize how the network prioritizes neighborhood information across different graph datasets. Our analysis of the learned architectures reveals that neighborhood mixing varies per datasets.'
volume: 97
URL: https://proceedings.mlr.press/v97/abu-el-haija19a.html
PDF: http://proceedings.mlr.press/v97/abu-el-haija19a/abu-el-haija19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-abu-el-haija19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sami
family: Abu-El-Haija
- given: Bryan
family: Perozzi
- given: Amol
family: Kapoor
- given: Nazanin
family: Alipourfard
- given: Kristina
family: Lerman
- given: Hrayr
family: Harutyunyan
- given: Greg Ver
family: Steeg
- given: Aram
family: Galstyan
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 21-29
id: abu-el-haija19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 21
lastpage: 29
published: 2019-05-24 00:00:00 +0000
- title: 'Communication-Constrained Inference and the Role of Shared Randomness'
abstract: 'A central server needs to perform statistical inference based on samples that are distributed over multiple users who can each send a message of limited length to the center. We study problems of distribution learning and identity testing in this distributed inference setting and examine the role of shared randomness as a resource. We propose a general purpose *simulate-and-infer* strategy that uses only private-coin communication protocols and is sample-optimal for distribution learning. This general strategy turns out to be sample-optimal even for distribution testing among private-coin protocols. Interestingly, we propose a public-coin protocol that outperforms simulate-and-infer for distribution testing and is, in fact, sample-optimal. Underlying our public-coin protocol is a random hash that when applied to the samples minimally contracts the chi-squared distance of their distribution from the uniform distribution.'
volume: 97
URL: https://proceedings.mlr.press/v97/acharya19a.html
PDF: http://proceedings.mlr.press/v97/acharya19a/acharya19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-acharya19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jayadev
family: Acharya
- given: Clement
family: Canonne
- given: Himanshu
family: Tyagi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 30-39
id: acharya19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 30
lastpage: 39
published: 2019-05-24 00:00:00 +0000
- title: 'Distributed Learning with Sublinear Communication'
abstract: 'In distributed statistical learning, $N$ samples are split across $m$ machines and a learner wishes to use minimal communication to learn as well as if the examples were on a single machine. This model has received substantial interest in machine learning due to its scalability and potential for parallel speedup. However, in high-dimensional settings, where the number examples is smaller than the number of features (‘"dimension"), the speedup afforded by distributed learning may be overshadowed by the cost of communicating a single example. This paper investigates the following question: When is it possible to learn a $d$-dimensional model in the distributed setting with total communication sublinear in $d$? Starting with a negative result, we observe that for learning $\ell_1$-bounded or sparse linear models, no algorithm can obtain optimal error until communication is linear in dimension. Our main result is that by slightly relaxing the standard boundedness assumptions for linear models, we can obtain distributed algorithms that enjoy optimal error with communication *logarithmic* in dimension. This result is based on a family of algorithms that combine mirror descent with randomized sparsification/quantization of iterates, and extends to the general stochastic convex optimization model.'
volume: 97
URL: https://proceedings.mlr.press/v97/acharya19b.html
PDF: http://proceedings.mlr.press/v97/acharya19b/acharya19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-acharya19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jayadev
family: Acharya
- given: Chris
family: De Sa
- given: Dylan
family: Foster
- given: Karthik
family: Sridharan
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 40-50
id: acharya19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 40
lastpage: 50
published: 2019-05-24 00:00:00 +0000
- title: 'Communication Complexity in Locally Private Distribution Estimation and Heavy Hitters'
abstract: 'We consider the problems of distribution estimation, and heavy hitter (frequency) estimation under privacy, and communication constraints. While the constraints have been studied separately, optimal schemes for one are sub-optimal for the other. We propose a sample-optimal $\eps$-locally differentially private (LDP) scheme for distribution estimation, where each user communicates one bit, and requires *no* public randomness. We also show that Hadamard Response, a recently proposed scheme for $\eps$-LDP distribution estimation is also utility-optimal for heavy hitters estimation. Our final result shows that unlike distribution estimation, without public randomness, any utility-optimal heavy hitter estimation algorithm must require $\Omega(\log n)$ bits of communication per user.'
volume: 97
URL: https://proceedings.mlr.press/v97/acharya19c.html
PDF: http://proceedings.mlr.press/v97/acharya19c/acharya19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-acharya19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jayadev
family: Acharya
- given: Ziteng
family: Sun
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 51-60
id: acharya19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 51
lastpage: 60
published: 2019-05-24 00:00:00 +0000
- title: 'Learning Models from Data with Measurement Error: Tackling Underreporting'
abstract: 'Measurement error in observational datasets can lead to systematic bias in inferences based on these datasets. As studies based on observational data are increasingly used to inform decisions with real-world impact, it is critical that we develop a robust set of techniques for analyzing and adjusting for these biases. In this paper we present a method for estimating the distribution of an outcome given a binary exposure that is subject to underreporting. Our method is based on a missing data view of the measurement error problem, where the true exposure is treated as a latent variable that is marginalized out of a joint model. We prove three different conditions under which the outcome distribution can still be identified from data containing only error-prone observations of the exposure. We demonstrate this method on synthetic data and analyze its sensitivity to near violations of the identifiability conditions. Finally, we use this method to estimate the effects of maternal smoking and heroin use during pregnancy on childhood obesity, two import problems from public health. Using the proposed method, we estimate these effects using only subject-reported drug use data and refine the range of estimates generated by a sensitivity analysis-based approach. Further, the estimates produced by our method are consistent with existing literature on both the effects of maternal smoking and the rate at which subjects underreport smoking.'
volume: 97
URL: https://proceedings.mlr.press/v97/adams19a.html
PDF: http://proceedings.mlr.press/v97/adams19a/adams19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-adams19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Roy
family: Adams
- given: Yuelong
family: Ji
- given: Xiaobin
family: Wang
- given: Suchi
family: Saria
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 61-70
id: adams19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 61
lastpage: 70
published: 2019-05-24 00:00:00 +0000
- title: 'TibGM: A Transferable and Information-Based Graphical Model Approach for Reinforcement Learning'
abstract: 'One of the challenges to reinforcement learning (RL) is scalable transferability among complex tasks. Incorporating a graphical model (GM), along with the rich family of related methods, as a basis for RL frameworks provides potential to address issues such as transferability, generalisation and exploration. Here we propose a flexible GM-based RL framework which leverages efficient inference procedures to enhance generalisation and transfer power. In our proposed transferable and information-based graphical model framework ‘TibGM’, we show the equivalence between our mutual information-based objective in the GM, and an RL consolidated objective consisting of a standard reward maximisation target and a generalisation/transfer objective. In settings where there is a sparse or deceptive reward signal, our TibGM framework is flexible enough to incorporate exploration bonuses depicting intrinsic rewards. We empirically verify improved performance and exploration power.'
volume: 97
URL: https://proceedings.mlr.press/v97/adel19a.html
PDF: http://proceedings.mlr.press/v97/adel19a/adel19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-adel19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Tameem
family: Adel
- given: Adrian
family: Weller
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 71-81
id: adel19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 71
lastpage: 81
published: 2019-05-24 00:00:00 +0000
- title: 'PAC Learnability of Node Functions in Networked Dynamical Systems'
abstract: 'We consider the PAC learnability of the local functions at the vertices of a discrete networked dynamical system, assuming that the underlying network is known. Our focus is on the learnability of threshold functions. We show that several variants of threshold functions are PAC learnable and provide tight bounds on the sample complexity. In general, when the input consists of positive and negative examples, we show that the concept class of threshold functions is not efficiently PAC learnable, unless NP = RP. Using a dynamic programming approach, we show efficient PAC learnability when the number of negative examples is small. We also present an efficient learner which is consistent with all the positive examples and at least (1-1/e) fraction of the negative examples. This algorithm is based on maximizing a submodular function under matroid constraints. By performing experiments on both synthetic and real-world networks, we study how the network structure and sample complexity influence the quality of the inferred system.'
volume: 97
URL: https://proceedings.mlr.press/v97/adiga19a.html
PDF: http://proceedings.mlr.press/v97/adiga19a/adiga19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-adiga19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Abhijin
family: Adiga
- given: Chris J
family: Kuhlman
- given: Madhav
family: Marathe
- given: S
family: Ravi
- given: Anil
family: Vullikanti
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 82-91
id: adiga19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 82
lastpage: 91
published: 2019-05-24 00:00:00 +0000
- title: 'Static Automatic Batching In TensorFlow'
abstract: 'Dynamic neural networks are becoming increasingly common, and yet it is hard to implement them efficiently. On-the-fly operation batching for such models is sub-optimal and suffers from run time overheads, while writing manually batched versions can be hard and error-prone. To address this we extend TensorFlow with pfor, a parallel-for loop optimized using static loop vectorization. With pfor, users can express computation using nested loops and conditional constructs, but get performance resembling that of a manually batched version. Benchmarks demonstrate speedups of one to two orders of magnitude on range of tasks, from jacobian computation, to Graph Neural Networks.'
volume: 97
URL: https://proceedings.mlr.press/v97/agarwal19a.html
PDF: http://proceedings.mlr.press/v97/agarwal19a/agarwal19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-agarwal19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ashish
family: Agarwal
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 92-101
id: agarwal19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 92
lastpage: 101
published: 2019-05-24 00:00:00 +0000
- title: 'Efficient Full-Matrix Adaptive Regularization'
abstract: 'Adaptive regularization methods pre-multiply a descent direction by a preconditioning matrix. Due to the large number of parameters of machine learning problems, full-matrix preconditioning methods are prohibitively expensive. We show how to modify full-matrix adaptive regularization in order to make it practical and effective. We also provide a novel theoretical analysis for adaptive regularization in *non-convex* optimization settings. The core of our algorithm, termed GGT, consists of the efficient computation of the inverse square root of a low-rank matrix. Our preliminary experiments show improved iteration-wise convergence rates across synthetic tasks and standard deep learning benchmarks, and that the more carefully-preconditioned steps sometimes lead to a better solution.'
volume: 97
URL: https://proceedings.mlr.press/v97/agarwal19b.html
PDF: http://proceedings.mlr.press/v97/agarwal19b/agarwal19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-agarwal19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Naman
family: Agarwal
- given: Brian
family: Bullins
- given: Xinyi
family: Chen
- given: Elad
family: Hazan
- given: Karan
family: Singh
- given: Cyril
family: Zhang
- given: Yi
family: Zhang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 102-110
id: agarwal19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 102
lastpage: 110
published: 2019-05-24 00:00:00 +0000
- title: 'Online Control with Adversarial Disturbances'
abstract: 'We study the control of linear dynamical systems with adversarial disturbances, as opposed to statistical noise. We present an efficient algorithm that achieves nearly-tight regret bounds in this setting. Our result generalizes upon previous work in two main aspects: the algorithm can accommodate adversarial noise in the dynamics, and can handle general convex costs.'
volume: 97
URL: https://proceedings.mlr.press/v97/agarwal19c.html
PDF: http://proceedings.mlr.press/v97/agarwal19c/agarwal19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-agarwal19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Naman
family: Agarwal
- given: Brian
family: Bullins
- given: Elad
family: Hazan
- given: Sham
family: Kakade
- given: Karan
family: Singh
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 111-119
id: agarwal19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 111
lastpage: 119
published: 2019-05-24 00:00:00 +0000
- title: 'Fair Regression: Quantitative Definitions and Reduction-Based Algorithms'
abstract: 'In this paper, we study the prediction of a real-valued target, such as a risk score or recidivism rate, while guaranteeing a quantitative notion of fairness with respect to a protected attribute such as gender or race. We call this class of problems fair regression. We propose general schemes for fair regression under two notions of fairness: (1) statistical parity, which asks that the prediction be statistically independent of the protected attribute, and (2) bounded group loss, which asks that the prediction error restricted to any protected group remain below some pre-determined level. While we only study these two notions of fairness, our schemes are applicable to arbitrary Lipschitz-continuous losses, and so they encompass least-squares regression, logistic regression, quantile regression, and many other tasks. Our schemes only require access to standard risk minimization algorithms (such as standard classification or least-squares regression) while providing theoretical guarantees on the optimality and fairness of the obtained solutions. In addition to analyzing theoretical properties of our schemes, we empirically demonstrate their ability to uncover fairness–accuracy frontiers on several standard datasets.'
volume: 97
URL: https://proceedings.mlr.press/v97/agarwal19d.html
PDF: http://proceedings.mlr.press/v97/agarwal19d/agarwal19d.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-agarwal19d.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alekh
family: Agarwal
- given: Miroslav
family: Dudik
- given: Zhiwei Steven
family: Wu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 120-129
id: agarwal19d
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 120
lastpage: 129
published: 2019-05-24 00:00:00 +0000
- title: 'Learning to Generalize from Sparse and Underspecified Rewards'
abstract: 'We consider the problem of learning from sparse and underspecified rewards, where an agent receives a complex input, such as a natural language instruction, and needs to generate a complex response, such as an action sequence, while only receiving binary success-failure feedback. Such success-failure rewards are often underspecified: they do not distinguish between purposeful and accidental success. Generalization from underspecified rewards hinges on discounting spurious trajectories that attain accidental success, while learning from sparse feedback requires effective exploration. We address exploration by using a mode covering direction of KL divergence to collect a diverse set of successful trajectories, followed by a mode seeking KL divergence to train a robust policy. We propose Meta Reward Learning (MeRL) to construct an auxiliary reward function that provides more refined feedback for learning. The parameters of the auxiliary reward function are optimized with respect to the validation performance of a trained policy. The MeRL approach outperforms an alternative method for reward learning based on Bayesian Optimization, and achieves the state-of-the-art on weakly-supervised semantic parsing. It improves previous work by 1.2% and 2.4% on WikiTableQuestions and WikiSQL datasets respectively.'
volume: 97
URL: https://proceedings.mlr.press/v97/agarwal19e.html
PDF: http://proceedings.mlr.press/v97/agarwal19e/agarwal19e.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-agarwal19e.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Rishabh
family: Agarwal
- given: Chen
family: Liang
- given: Dale
family: Schuurmans
- given: Mohammad
family: Norouzi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 130-140
id: agarwal19e
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 130
lastpage: 140
published: 2019-05-24 00:00:00 +0000
- title: 'The Kernel Interaction Trick: Fast Bayesian Discovery of Pairwise Interactions in High Dimensions'
abstract: 'Discovering interaction effects on a response of interest is a fundamental problem faced in biology, medicine, economics, and many other scientific disciplines. In theory, Bayesian methods for discovering pairwise interactions enjoy many benefits such as coherent uncertainty quantification, the ability to incorporate background knowledge, and desirable shrinkage properties. In practice, however, Bayesian methods are often computationally intractable for even moderate- dimensional problems. Our key insight is that many hierarchical models of practical interest admit a Gaussian process representation such that rather than maintaining a posterior over all O(p^2) interactions, we need only maintain a vector of O(p) kernel hyper-parameters. This implicit representation allows us to run Markov chain Monte Carlo (MCMC) over model hyper-parameters in time and memory linear in p per iteration. We focus on sparsity-inducing models and show on datasets with a variety of covariate behaviors that our method: (1) reduces runtime by orders of magnitude over naive applications of MCMC, (2) provides lower Type I and Type II error relative to state-of-the-art LASSO-based approaches, and (3) offers improved computational scaling in high dimensions relative to existing Bayesian and LASSO-based approaches.'
volume: 97
URL: https://proceedings.mlr.press/v97/agrawal19a.html
PDF: http://proceedings.mlr.press/v97/agrawal19a/agrawal19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-agrawal19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Raj
family: Agrawal
- given: Brian
family: Trippe
- given: Jonathan
family: Huggins
- given: Tamara
family: Broderick
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 141-150
id: agrawal19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 141
lastpage: 150
published: 2019-05-24 00:00:00 +0000
- title: 'Understanding the Impact of Entropy on Policy Optimization'
abstract: 'Entropy regularization is commonly used to improve policy optimization in reinforcement learning. It is believed to help with exploration by encouraging the selection of more stochastic policies. In this work, we analyze this claim using new visualizations of the optimization landscape based on randomly perturbing the loss function. We first show that even with access to the exact gradient, policy optimization is difficult due to the geometry of the objective function. We then qualitatively show that in some environments, a policy with higher entropy can make the optimization landscape smoother, thereby connecting local optima and enabling the use of larger learning rates. This paper presents new tools for understanding the optimization landscape, shows that policy entropy serves as a regularizer, and highlights the challenge of designing general-purpose policy optimization algorithms.'
volume: 97
URL: https://proceedings.mlr.press/v97/ahmed19a.html
PDF: http://proceedings.mlr.press/v97/ahmed19a/ahmed19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ahmed19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Zafarali
family: Ahmed
- given: Nicolas Le
family: Roux
- given: Mohammad
family: Norouzi
- given: Dale
family: Schuurmans
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 151-160
id: ahmed19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 151
lastpage: 160
published: 2019-05-24 00:00:00 +0000
- title: 'Fairwashing: the risk of rationalization'
abstract: 'Black-box explanation is the problem of explaining how a machine learning model – whose internal logic is hidden to the auditor and generally complex – produces its outcomes. Current approaches for solving this problem include model explanation, outcome explanation as well as model inspection. While these techniques can be beneficial by providing interpretability, they can be used in a negative manner to perform fairwashing, which we define as promoting the false perception that a machine learning model respects some ethical values. In particular, we demonstrate that it is possible to systematically rationalize decisions taken by an unfair black-box model using the model explanation as well as the outcome explanation approaches with a given fairness metric. Our solution, LaundryML, is based on a regularized rule list enumeration algorithm whose objective is to search for fair rule lists approximating an unfair black-box model. We empirically evaluate our rationalization technique on black-box models trained on real-world datasets and show that one can obtain rule lists with high fidelity to the black-box model while being considerably less unfair at the same time.'
volume: 97
URL: https://proceedings.mlr.press/v97/aivodji19a.html
PDF: http://proceedings.mlr.press/v97/aivodji19a/aivodji19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-aivodji19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ulrich
family: Aivodji
- given: Hiromi
family: Arai
- given: Olivier
family: Fortineau
- given: Sébastien
family: Gambs
- given: Satoshi
family: Hara
- given: Alain
family: Tapp
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 161-170
id: aivodji19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 161
lastpage: 170
published: 2019-05-24 00:00:00 +0000
- title: 'Adaptive Stochastic Natural Gradient Method for One-Shot Neural Architecture Search'
abstract: 'High sensitivity of neural architecture search (NAS) methods against their input such as step-size (i.e., learning rate) and search space prevents practitioners from applying them out-of-the-box to their own problems, albeit its purpose is to automate a part of tuning process. Aiming at a fast, robust, and widely-applicable NAS, we develop a generic optimization framework for NAS. We turn a coupled optimization of connection weights and neural architecture into a differentiable optimization by means of stochastic relaxation. It accepts arbitrary search space (widely-applicable) and enables to employ a gradient-based simultaneous optimization of weights and architecture (fast). We propose a stochastic natural gradient method with an adaptive step-size mechanism built upon our theoretical investigation (robust). Despite its simplicity and no problem-dependent parameter tuning, our method exhibited near state-of-the-art performances with low computational budgets both on image classification and inpainting tasks.'
volume: 97
URL: https://proceedings.mlr.press/v97/akimoto19a.html
PDF: http://proceedings.mlr.press/v97/akimoto19a/akimoto19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-akimoto19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Youhei
family: Akimoto
- given: Shinichi
family: Shirakawa
- given: Nozomu
family: Yoshinari
- given: Kento
family: Uchida
- given: Shota
family: Saito
- given: Kouhei
family: Nishida
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 171-180
id: akimoto19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 171
lastpage: 180
published: 2019-05-24 00:00:00 +0000
- title: 'Projections for Approximate Policy Iteration Algorithms'
abstract: 'Approximate policy iteration is a class of reinforcement learning (RL) algorithms where the policy is encoded using a function approximator and which has been especially prominent in RL with continuous action spaces. In this class of RL algorithms, ensuring increase of the policy return during policy update often requires to constrain the change in action distribution. Several approximations exist in the literature to solve this constrained policy update problem. In this paper, we propose to improve over such solutions by introducing a set of projections that transform the constrained problem into an unconstrained one which is then solved by standard gradient descent. Using these projections, we empirically demonstrate that our approach can improve the policy update solution and the control over exploration of existing approximate policy iteration algorithms.'
volume: 97
URL: https://proceedings.mlr.press/v97/akrour19a.html
PDF: http://proceedings.mlr.press/v97/akrour19a/akrour19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-akrour19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Riad
family: Akrour
- given: Joni
family: Pajarinen
- given: Jan
family: Peters
- given: Gerhard
family: Neumann
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 181-190
id: akrour19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 181
lastpage: 190
published: 2019-05-24 00:00:00 +0000
- title: 'Validating Causal Inference Models via Influence Functions'
abstract: 'The problem of estimating causal effects of treatments from observational data falls beyond the realm of supervised learning {—} because counterfactual data is inaccessible, we can never observe the true causal effects. In the absence of "supervision", how can we evaluate the performance of causal inference methods? In this paper, we use influence functions {—} the functional derivatives of a loss function {—} to develop a model validation procedure that estimates the estimation error of causal inference methods. Our procedure utilizes a Taylor-like expansion to approximate the loss function of a method on a given dataset in terms of the influence functions of its loss on a "synthesized", proximal dataset with known causal effects. Under minimal regularity assumptions, we show that our procedure is consistent and efficient. Experiments on 77 benchmark datasets show that using our procedure, we can accurately predict the comparative performances of state-of-the-art causal inference methods applied to a given observational study.'
volume: 97
URL: https://proceedings.mlr.press/v97/alaa19a.html
PDF: http://proceedings.mlr.press/v97/alaa19a/alaa19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-alaa19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ahmed
family: Alaa
- given: Mihaela
family: Van Der Schaar
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 191-201
id: alaa19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 191
lastpage: 201
published: 2019-05-24 00:00:00 +0000
- title: 'Multi-objective training of Generative Adversarial Networks with multiple discriminators'
abstract: 'Recent literature has demonstrated promising results for training Generative Adversarial Networks by employing a set of discriminators, in contrast to the traditional game involving one generator against a single adversary. Such methods perform single-objective optimization on some simple consolidation of the losses, e.g. an arithmetic average. In this work, we revisit the multiple-discriminator setting by framing the simultaneous minimization of losses provided by different models as a multi-objective optimization problem. Specifically, we evaluate the performance of multiple gradient descent and the hypervolume maximization algorithm on a number of different datasets. Moreover, we argue that the previously proposed methods and hypervolume maximization can all be seen as variations of multiple gradient descent in which the update direction can be computed efficiently. Our results indicate that hypervolume maximization presents a better compromise between sample quality and computational cost than previous methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/albuquerque19a.html
PDF: http://proceedings.mlr.press/v97/albuquerque19a/albuquerque19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-albuquerque19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Isabela
family: Albuquerque
- given: Joao
family: Monteiro
- given: Thang
family: Doan
- given: Breandan
family: Considine
- given: Tiago
family: Falk
- given: Ioannis
family: Mitliagkas
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 202-211
id: albuquerque19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 202
lastpage: 211
published: 2019-05-24 00:00:00 +0000
- title: 'Graph Element Networks: adaptive, structured computation and memory'
abstract: 'We explore the use of graph neural networks (GNNs) to model spatial processes in which there is no a priori graphical structure. Similar to finite element analysis, we assign nodes of a GNN to spatial locations and use a computational process defined on the graph to model the relationship between an initial function defined over a space and a resulting function in the same space. We use GNNs as a computational substrate, and show that the locations of the nodes in space as well as their connectivity can be optimized to focus on the most complex parts of the space. Moreover, this representational strategy allows the learned input-output relationship to generalize over the size of the underlying space and run the same model at different levels of precision, trading computation for accuracy. We demonstrate this method on a traditional PDE problem, a physical prediction problem from robotics, and learning to predict scene images from novel viewpoints.'
volume: 97
URL: https://proceedings.mlr.press/v97/alet19a.html
PDF: http://proceedings.mlr.press/v97/alet19a/alet19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-alet19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ferran
family: Alet
- given: Adarsh Keshav
family: Jeewajee
- given: Maria Bauza
family: Villalonga
- given: Alberto
family: Rodriguez
- given: Tomas
family: Lozano-Perez
- given: Leslie
family: Kaelbling
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 212-222
id: alet19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 212
lastpage: 222
published: 2019-05-24 00:00:00 +0000
- title: 'Analogies Explained: Towards Understanding Word Embeddings'
abstract: 'Word embeddings generated by neural network methods such as word2vec (W2V) are well known to exhibit seemingly linear behaviour, e.g. the embeddings of analogy “woman is to queen as man is to king” approximately describe a parallelogram. This property is particularly intriguing since the embeddings are not trained to achieve it. Several explanations have been proposed, but each introduces assumptions that do not hold in practice. We derive a probabilistically grounded definition of paraphrasing that we re-interpret as word transformation, a mathematical description of “$w_x$ is to $w_y$”. From these concepts we prove existence of linear relationship between W2V-type embeddings that underlie the analogical phenomenon, identifying explicit error terms.'
volume: 97
URL: https://proceedings.mlr.press/v97/allen19a.html
PDF: http://proceedings.mlr.press/v97/allen19a/allen19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-allen19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Carl
family: Allen
- given: Timothy
family: Hospedales
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 223-231
id: allen19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 223
lastpage: 231
published: 2019-05-24 00:00:00 +0000
- title: 'Infinite Mixture Prototypes for Few-shot Learning'
abstract: 'We propose infinite mixture prototypes to adaptively represent both simple and complex data distributions for few-shot learning. Infinite mixture prototypes combine deep representation learning with Bayesian nonparametrics, representing each class by a set of clusters, unlike existing prototypical methods that represent each class by a single cluster. By inferring the number of clusters, infinite mixture prototypes interpolate between nearest neighbor and prototypical representations in a learned feature space, which improves accuracy and robustness in the few-shot regime. We show the importance of adaptive capacity for capturing complex data distributions such as super-classes (like alphabets in character recognition), with 10-25% absolute accuracy improvements over prototypical networks, while still maintaining or improving accuracy on standard few-shot learning benchmarks. By clustering labeled and unlabeled data with the same rule, infinite mixture prototypes achieve state-of-the-art semi-supervised accuracy, and can perform purely unsupervised clustering, unlike existing fully- and semi-supervised prototypical methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/allen19b.html
PDF: http://proceedings.mlr.press/v97/allen19b/allen19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-allen19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Kelsey
family: Allen
- given: Evan
family: Shelhamer
- given: Hanul
family: Shin
- given: Joshua
family: Tenenbaum
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 232-241
id: allen19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 232
lastpage: 241
published: 2019-05-24 00:00:00 +0000
- title: 'A Convergence Theory for Deep Learning via Over-Parameterization'
abstract: 'Deep neural networks (DNNs) have demonstrated dominating performance in many fields; since AlexNet, networks used in practice are going wider and deeper. On the theoretical side, a long line of works have been focusing on why we can train neural networks when there is only one hidden layer. The theory of multi-layer networks remains unsettled. In this work, we prove simple algorithms such as stochastic gradient descent (SGD) can find Global Minima on the training objective of DNNs in Polynomial Time. We only make two assumptions: the inputs do not degenerate and the network is over-parameterized. The latter means the number of hidden neurons is sufficiently large: polynomial in L, the number of DNN layers and in n, the number of training samples. As concrete examples, starting from randomly initialized weights, we show that SGD attains 100% training accuracy in classification tasks, or minimizes regression loss in linear convergence speed eps e^{-T}, with running time polynomial in n and L. Our theory applies to the widely-used but non-smooth ReLU activation, and to any smooth and possibly non-convex loss functions. In terms of network architectures, our theory at least applies to fully-connected neural networks, convolutional neural networks (CNN), and residual neural networks (ResNet).'
volume: 97
URL: https://proceedings.mlr.press/v97/allen-zhu19a.html
PDF: http://proceedings.mlr.press/v97/allen-zhu19a/allen-zhu19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-allen-zhu19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Zeyuan
family: Allen-Zhu
- given: Yuanzhi
family: Li
- given: Zhao
family: Song
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 242-252
id: allen-zhu19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 242
lastpage: 252
published: 2019-05-24 00:00:00 +0000
- title: 'Asynchronous Batch Bayesian Optimisation with Improved Local Penalisation'
abstract: 'Batch Bayesian optimisation (BO) has been successfully applied to hyperparameter tuning using parallel computing, but it is wasteful of resources: workers that complete jobs ahead of others are left idle. We address this problem by developing an approach, Penalising Locally for Asynchronous Bayesian Optimisation on K Workers (PLAyBOOK), for asynchronous parallel BO. We demonstrate empirically the efficacy of PLAyBOOK and its variants on synthetic tasks and a real-world problem. We undertake a comparison between synchronous and asynchronous BO, and show that asynchronous BO often outperforms synchronous batch BO in both wall-clock time and sample efficiency.'
volume: 97
URL: https://proceedings.mlr.press/v97/alvi19a.html
PDF: http://proceedings.mlr.press/v97/alvi19a/alvi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-alvi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ahsan
family: Alvi
- given: Binxin
family: Ru
- given: Jan-Peter
family: Calliess
- given: Stephen
family: Roberts
- given: Michael A.
family: Osborne
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 253-262
id: alvi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 253
lastpage: 262
published: 2019-05-24 00:00:00 +0000
- title: 'Bounding User Contributions: A Bias-Variance Trade-off in Differential Privacy'
abstract: 'Differentially private learning algorithms protect individual participants in the training dataset by guaranteeing that their presence does not significantly change the resulting model. In order to make this promise, such algorithms need to know the maximum contribution that can be made by a single user: the more data an individual can contribute, the more noise will need to be added to protect them. While most existing analyses assume that the maximum contribution is known and fixed in advance{—}indeed, it is often assumed that each user contributes only a single example{—}we argue that in practice there is a meaningful choice to be made. On the one hand, if we allow users to contribute large amounts of data, we may end up adding excessive noise to protect a few outliers, even when the majority contribute only modestly. On the other hand, limiting users to small contributions keeps noise levels low at the cost of potentially discarding significant amounts of excess data, thus introducing bias. Here, we characterize this trade-off for an empirical risk minimization setting, showing that in general there is a “sweet spot” that depends on measurable properties of the dataset, but that there is also a concrete cost to privacy that cannot be avoided simply by collecting more data.'
volume: 97
URL: https://proceedings.mlr.press/v97/amin19a.html
PDF: http://proceedings.mlr.press/v97/amin19a/amin19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-amin19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Kareem
family: Amin
- given: Alex
family: Kulesza
- given: Andres
family: Munoz
- given: Sergei
family: Vassilvtiskii
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 263-271
id: amin19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 263
lastpage: 271
published: 2019-05-24 00:00:00 +0000
- title: 'Explaining Deep Neural Networks with a Polynomial Time Algorithm for Shapley Value Approximation'
abstract: 'The problem of explaining the behavior of deep neural networks has recently gained a lot of attention. While several attribution methods have been proposed, most come without strong theoretical foundations, which raises questions about their reliability. On the other hand, the literature on cooperative game theory suggests Shapley values as a unique way of assigning relevance scores such that certain desirable properties are satisfied. Unfortunately, the exact evaluation of Shapley values is prohibitively expensive, exponential in the number of input features. In this work, by leveraging recent results on uncertainty propagation, we propose a novel, polynomial-time approximation of Shapley values in deep neural networks. We show that our method produces significantly better approximations of Shapley values than existing state-of-the-art attribution methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/ancona19a.html
PDF: http://proceedings.mlr.press/v97/ancona19a/ancona19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ancona19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Marco
family: Ancona
- given: Cengiz
family: Oztireli
- given: Markus
family: Gross
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 272-281
id: ancona19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 272
lastpage: 281
published: 2019-05-24 00:00:00 +0000
- title: 'Scaling Up Ordinal Embedding: A Landmark Approach'
abstract: 'Ordinal Embedding is the problem of placing n objects into R^d to satisfy constraints like "object a is closer to b than to c." It can accommodate data that embeddings from features or distances cannot, but is a more difficult problem. We propose a novel landmark-based method as a partial solution. At small to medium scales, we present a novel combination of existing methods with some new theoretical justification. For very large values of n optimizing over an entire embedding breaks down, so we propose a novel method which first embeds a subset of m << n objects and then embeds the remaining objects independently and in parallel. We prove a distance error bound for our method in terms of m and that it has O(dn log m) time complexity, and show empirically that it is able to produce high quality embeddings in a fraction of the time needed for any published method.'
volume: 97
URL: https://proceedings.mlr.press/v97/anderton19a.html
PDF: http://proceedings.mlr.press/v97/anderton19a/anderton19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-anderton19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jesse
family: Anderton
- given: Javed
family: Aslam
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 282-290
id: anderton19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 282
lastpage: 290
published: 2019-05-24 00:00:00 +0000
- title: 'Sorting Out Lipschitz Function Approximation'
abstract: 'Training neural networks under a strict Lipschitz constraint is useful for provable adversarial robustness, generalization bounds, interpretable gradients, and Wasserstein distance estimation. By the composition property of Lipschitz functions, it suffices to ensure that each individual affine transformation or nonlinear activation is 1-Lipschitz. The challenge is to do this while maintaining the expressive power. We identify a necessary property for such an architecture: each of the layers must preserve the gradient norm during backpropagation. Based on this, we propose to combine a gradient norm preserving activation function, GroupSort, with norm-constrained weight matrices. We show that norm-constrained GroupSort architectures are universal Lipschitz function approximators. Empirically, we show that norm-constrained GroupSort networks achieve tighter estimates of Wasserstein distance than their ReLU counterparts and can achieve provable adversarial robustness guarantees with little cost to accuracy.'
volume: 97
URL: https://proceedings.mlr.press/v97/anil19a.html
PDF: http://proceedings.mlr.press/v97/anil19a/anil19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-anil19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Cem
family: Anil
- given: James
family: Lucas
- given: Roger
family: Grosse
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 291-301
id: anil19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 291
lastpage: 301
published: 2019-05-24 00:00:00 +0000
- title: 'Sparse Multi-Channel Variational Autoencoder for the Joint Analysis of Heterogeneous Data'
abstract: 'Interpretable modeling of heterogeneous data channels is essential in medical applications, for example when jointly analyzing clinical scores and medical images. Variational Autoencoders (VAE) are powerful generative models that learn representations of complex data. The flexibility of VAE may come at the expense of lack of interpretability in describing the joint relationship between heterogeneous data. To tackle this problem, in this work we extend the variational framework of VAE to bring parsimony and interpretability when jointly account for latent relationships across multiple channels. In the latent space, this is achieved by constraining the variational distribution of each channel to a common target prior. Parsimonious latent representations are enforced by variational dropout. Experiments on synthetic data show that our model correctly identifies the prescribed latent dimensions and data relationships across multiple testing scenarios. When applied to imaging and clinical data, our method allows to identify the joint effect of age and pathology in describing clinical condition in a large scale clinical cohort.'
volume: 97
URL: https://proceedings.mlr.press/v97/antelmi19a.html
PDF: http://proceedings.mlr.press/v97/antelmi19a/antelmi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-antelmi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Luigi
family: Antelmi
- given: Nicholas
family: Ayache
- given: Philippe
family: Robert
- given: Marco
family: Lorenzi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 302-311
id: antelmi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 302
lastpage: 311
published: 2019-05-24 00:00:00 +0000
- title: 'Unsupervised Label Noise Modeling and Loss Correction'
abstract: 'Despite being robust to small amounts of label noise, convolutional neural networks trained with stochastic gradient methods have been shown to easily fit random labels. When there are a mixture of correct and mislabelled targets, networks tend to fit the former before the latter. This suggests using a suitable two-component mixture model as an unsupervised generative model of sample loss values during training to allow online estimation of the probability that a sample is mislabelled. Specifically, we propose a beta mixture to estimate this probability and correct the loss by relying on the network prediction (the so-called bootstrapping loss). We further adapt mixup augmentation to drive our approach a step further. Experiments on CIFAR-10/100 and TinyImageNet demonstrate a robustness to label noise that substantially outperforms recent state-of-the-art. Source code is available at https://git.io/fjsvE and Appendix at https://arxiv.org/abs/1904.11238.'
volume: 97
URL: https://proceedings.mlr.press/v97/arazo19a.html
PDF: http://proceedings.mlr.press/v97/arazo19a/arazo19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-arazo19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Eric
family: Arazo
- given: Diego
family: Ortego
- given: Paul
family: Albert
- given: Noel
family: O’Connor
- given: Kevin
family: Mcguinness
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 312-321
id: arazo19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 312
lastpage: 321
published: 2019-05-24 00:00:00 +0000
- title: 'Fine-Grained Analysis of Optimization and Generalization for Overparameterized Two-Layer Neural Networks'
abstract: 'Recent works have cast some light on the mystery of why deep nets fit any data and generalize despite being very overparametrized. This paper analyzes training and generalization for a simple 2-layer ReLU net with random initialization, and provides the following improvements over recent works: (i) Using a tighter characterization of training speed than recent papers, an explanation for why training a neural net with random labels leads to slower training, as originally observed in [Zhang et al. ICLR’17]. (ii) Generalization bound independent of network size, using a data-dependent complexity measure. Our measure distinguishes clearly between random labels and true labels on MNIST and CIFAR, as shown by experiments. Moreover, recent papers require sample complexity to increase (slowly) with the size, while our sample complexity is completely independent of the network size. (iii) Learnability of a broad class of smooth functions by 2-layer ReLU nets trained via gradient descent. The key idea is to track dynamics of training and generalization via properties of a related kernel.'
volume: 97
URL: https://proceedings.mlr.press/v97/arora19a.html
PDF: http://proceedings.mlr.press/v97/arora19a/arora19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-arora19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sanjeev
family: Arora
- given: Simon
family: Du
- given: Wei
family: Hu
- given: Zhiyuan
family: Li
- given: Ruosong
family: Wang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 322-332
id: arora19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 322
lastpage: 332
published: 2019-05-24 00:00:00 +0000
- title: 'Distributed Weighted Matching via Randomized Composable Coresets'
abstract: 'Maximum weight matching is one of the most fundamental combinatorial optimization problems with a wide range of applications in data mining and bioinformatics. Developing distributed weighted matching algorithms has been challenging due to the sequential nature of efficient algorithms for this problem. In this paper, we develop a simple distributed algorithm for the problem on general graphs with approximation guarantee of 2 + eps that (nearly) matches that of the sequential greedy algorithm. A key advantage of this algorithm is that it can be easily implemented in only two rounds of computation in modern parallel computation frameworks such as MapReduce. We also demonstrate the efficiency of our algorithm in practice on various graphs (some with half a trillion edges) by achieving objective values always close to what is achievable in the centralized setting.'
volume: 97
URL: https://proceedings.mlr.press/v97/assadi19a.html
PDF: http://proceedings.mlr.press/v97/assadi19a/assadi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-assadi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sepehr
family: Assadi
- given: Mohammadhossein
family: Bateni
- given: Vahab
family: Mirrokni
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 333-343
id: assadi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 333
lastpage: 343
published: 2019-05-24 00:00:00 +0000
- title: 'Stochastic Gradient Push for Distributed Deep Learning'
abstract: 'Distributed data-parallel algorithms aim to accelerate the training of deep neural networks by parallelizing the computation of large mini-batch gradient updates across multiple nodes. Approaches that synchronize nodes using exact distributed averaging (e.g., via AllReduce) are sensitive to stragglers and communication delays. The PushSum gossip algorithm is robust to these issues, but only performs approximate distributed averaging. This paper studies Stochastic Gradient Push (SGP), which combines PushSum with stochastic gradient updates. We prove that SGP converges to a stationary point of smooth, non-convex objectives at the same sub-linear rate as SGD, and that all nodes achieve consensus. We empirically validate the performance of SGP on image classification (ResNet-50, ImageNet) and machine translation (Transformer, WMT’16 En-De) workloads.'
volume: 97
URL: https://proceedings.mlr.press/v97/assran19a.html
PDF: http://proceedings.mlr.press/v97/assran19a/assran19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-assran19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Mahmoud
family: Assran
- given: Nicolas
family: Loizou
- given: Nicolas
family: Ballas
- given: Mike
family: Rabbat
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 344-353
id: assran19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 344
lastpage: 353
published: 2019-05-24 00:00:00 +0000
- title: 'Bayesian Optimization of Composite Functions'
abstract: 'We consider optimization of composite objective functions, i.e., of the form $f(x)=g(h(x))$, where $h$ is a black-box derivative-free expensive-to-evaluate function with vector-valued outputs, and $g$ is a cheap-to-evaluate real-valued function. While these problems can be solved with standard Bayesian optimization, we propose a novel approach that exploits the composite structure of the objective function to substantially improve sampling efficiency. Our approach models $h$ using a multi-output Gaussian process and chooses where to sample using the expected improvement evaluated on the implied non-Gaussian posterior on $f$, which we call expected improvement for composite functions (EI-CF). Although EI-CF cannot be computed in closed form, we provide a novel stochastic gradient estimator that allows its efficient maximization. We also show that our approach is asymptotically consistent, i.e., that it recovers a globally optimal solution as sampling effort grows to infinity, generalizing previous convergence results for classical expected improvement. Numerical experiments show that our approach dramatically outperforms standard Bayesian optimization benchmarks, reducing simple regret by several orders of magnitude.'
volume: 97
URL: https://proceedings.mlr.press/v97/astudillo19a.html
PDF: http://proceedings.mlr.press/v97/astudillo19a/astudillo19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-astudillo19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Raul
family: Astudillo
- given: Peter
family: Frazier
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 354-363
id: astudillo19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 354
lastpage: 363
published: 2019-05-24 00:00:00 +0000
- title: 'Linear-Complexity Data-Parallel Earth Mover’s Distance Approximations'
abstract: 'The Earth Mover’s Distance (EMD) is a state-of-the art metric for comparing discrete probability distributions, but its high distinguishability comes at a high cost in computational complexity. Even though linear-complexity approximation algorithms have been proposed to improve its scalability, these algorithms are either limited to vector spaces with only a few dimensions or they become ineffective when the degree of overlap between the probability distributions is high. We propose novel approximation algorithms that overcome both of these limitations, yet still achieve linear time complexity. All our algorithms are data parallel, and therefore, we can take advantage of massively parallel computing engines, such as Graphics Processing Units (GPUs). On the popular text-based 20 Newsgroups dataset, the new algorithms are four orders of magnitude faster than a multi-threaded CPU implementation of Word Mover’s Distance and match its search accuracy. On MNIST images, the new algorithms are four orders of magnitude faster than Cuturi’s GPU implementation of the Sinkhorn’s algorithm while offering a slightly higher search accuracy.'
volume: 97
URL: https://proceedings.mlr.press/v97/atasu19a.html
PDF: http://proceedings.mlr.press/v97/atasu19a/atasu19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-atasu19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Kubilay
family: Atasu
- given: Thomas
family: Mittelholzer
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 364-373
id: atasu19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 364
lastpage: 373
published: 2019-05-24 00:00:00 +0000
- title: 'Benefits and Pitfalls of the Exponential Mechanism with Applications to Hilbert Spaces and Functional PCA'
abstract: 'The exponential mechanism is a fundamental tool of Differential Privacy (DP) due to its strong privacy guarantees and flexibility. We study its extension to settings with summaries based on infinite dimensional outputs such as with functional data analysis, shape analysis, and nonparametric statistics. We show that the mechanism must be designed with respect to a specific base measure over the output space, such as a Gaussian process. We provide a positive result that establishes a Central Limit Theorem for the exponential mechanism quite broadly. We also provide a negative result, showing that the magnitude of noise introduced for privacy is asymptotically non-negligible relative to the statistical estimation error. We develop an $\ep$-DP mechanism for functional principal component analysis, applicable in separable Hilbert spaces, and demonstrate its performance via simulations and applications to two datasets.'
volume: 97
URL: https://proceedings.mlr.press/v97/awan19a.html
PDF: http://proceedings.mlr.press/v97/awan19a/awan19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-awan19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jordan
family: Awan
- given: Ana
family: Kenney
- given: Matthew
family: Reimherr
- given: Aleksandra
family: Slavković
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 374-384
id: awan19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 374
lastpage: 384
published: 2019-05-24 00:00:00 +0000
- title: 'Feature Grouping as a Stochastic Regularizer for High-Dimensional Structured Data'
abstract: 'In many applications where collecting data is expensive, for example neuroscience or medical imaging, the sample size is typically small compared to the feature dimension. These datasets call for intelligent regularization that exploits known structure, such as correlations between the features arising from the measurement device. However, existing structured regularizers need specially crafted solvers, which are difficult to apply to complex models. We propose a new regularizer specifically designed to leverage structure in the data in a way that can be applied efficiently to complex models. Our approach relies on feature grouping, using a fast clustering algorithm inside a stochastic gradient descent loop: given a family of feature groupings that capture feature covariations, we randomly select these groups at each iteration. Experiments on two real-world datasets demonstrate that the proposed approach produces models that generalize better than those trained with conventional regularizers, and also improves convergence speed, and has a linear computational cost.'
volume: 97
URL: https://proceedings.mlr.press/v97/aydore19a.html
PDF: http://proceedings.mlr.press/v97/aydore19a/aydore19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-aydore19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sergul
family: Aydore
- given: Bertrand
family: Thirion
- given: Gael
family: Varoquaux
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 385-394
id: aydore19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 385
lastpage: 394
published: 2019-05-24 00:00:00 +0000
- title: 'Beyond the Chinese Restaurant and Pitman-Yor processes: Statistical Models with double power-law behavior'
abstract: 'Bayesian nonparametric approaches, in particular the Pitman-Yor process and the associated two-parameter Chinese Restaurant process, have been successfully used in applications where the data exhibit a power-law behavior. Examples include natural language processing, natural images or networks. There is also growing empirical evidence suggesting that some datasets exhibit a two-regime power-law behavior: one regime for small frequencies, and a second regime, with a different exponent, for high frequencies. In this paper, we introduce a class of completely random measures which are doubly regularly-varying. Contrary to the Pitman-Yor process, we show that when completely random measures in this class are normalized to obtain random probability measures and associated random partitions, such partitions exhibit a double power-law behavior. We present two general constructions and discuss in particular two models within this class: the beta prime process (Broderick et al. (2015, 2018) and a novel process called generalized BFRY process. We derive efficient Markov chain Monte Carlo algorithms to estimate the parameters of these models. Finally, we show that the proposed models provide a better fit than the Pitman-Yor process on various datasets.'
volume: 97
URL: https://proceedings.mlr.press/v97/ayed19a.html
PDF: http://proceedings.mlr.press/v97/ayed19a/ayed19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ayed19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Fadhel
family: Ayed
- given: Juho
family: Lee
- given: Francois
family: Caron
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 395-404
id: ayed19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 395
lastpage: 404
published: 2019-05-24 00:00:00 +0000
- title: 'Scalable Fair Clustering'
abstract: 'We study the fair variant of the classic k-median problem introduced by (Chierichetti et al., NeurIPS 2017) in which the points are colored, and the goal is to minimize the same average distance objective as in the standard $k$-median problem while ensuring that all clusters have an “approximately equal” number of points of each color. (Chierichetti et al., NeurIPS 2017) proposed a two-phase algorithm for fair $k$-clustering. In the first step, the pointset is partitioned into subsets called fairlets that satisfy the fairness requirement and approximately preserve the k-median objective. In the second step, fairlets are merged into k clusters by one of the existing k-median algorithms. The running time of this algorithm is dominated by the first step, which takes super-quadratic time. In this paper, we present a practical approximate fairlet decomposition algorithm that runs in nearly linear time.'
volume: 97
URL: https://proceedings.mlr.press/v97/backurs19a.html
PDF: http://proceedings.mlr.press/v97/backurs19a/backurs19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-backurs19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Arturs
family: Backurs
- given: Piotr
family: Indyk
- given: Krzysztof
family: Onak
- given: Baruch
family: Schieber
- given: Ali
family: Vakilian
- given: Tal
family: Wagner
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 405-413
id: backurs19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 405
lastpage: 413
published: 2019-05-24 00:00:00 +0000
- title: 'Entropic GANs meet VAEs: A Statistical Approach to Compute Sample Likelihoods in GANs'
abstract: 'Building on the success of deep learning, two modern approaches to learn a probability model from the data are Generative Adversarial Networks (GANs) and Variational AutoEncoders (VAEs). VAEs consider an explicit probability model for the data and compute a generative distribution by maximizing a variational lower-bound on the log-likelihood function. GANs, however, compute a generative model by minimizing a distance between observed and generated probability distributions without considering an explicit model for the observed data. The lack of having explicit probability models in GANs prohibits computation of sample likelihoods in their frameworks and limits their use in statistical inference problems. In this work, we resolve this issue by constructing an explicit probability model that can be used to compute sample likelihood statistics in GANs. In particular, we prove that under this probability model, a family of Wasserstein GANs with an entropy regularization can be viewed as a generative model that maximizes a variational lower-bound on average sample log likelihoods, an approach that VAEs are based on. This result makes a principled connection between two modern generative models, namely GANs and VAEs. In addition to the aforementioned theoretical results, we compute likelihood statistics for GANs trained on Gaussian, MNIST, SVHN, CIFAR-10 and LSUN datasets. Our numerical results validate the proposed theory.'
volume: 97
URL: https://proceedings.mlr.press/v97/balaji19a.html
PDF: http://proceedings.mlr.press/v97/balaji19a/balaji19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-balaji19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yogesh
family: Balaji
- given: Hamed
family: Hassani
- given: Rama
family: Chellappa
- given: Soheil
family: Feizi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 414-423
id: balaji19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 414
lastpage: 423
published: 2019-05-24 00:00:00 +0000
- title: 'Provable Guarantees for Gradient-Based Meta-Learning'
abstract: 'We study the problem of meta-learning through the lens of online convex optimization, developing a meta-algorithm bridging the gap between popular gradient-based meta-learning and classical regularization-based multi-task transfer methods. Our method is the first to simultaneously satisfy good sample efficiency guarantees in the convex setting, with generalization bounds that improve with task-similarity, while also being computationally scalable to modern deep learning architectures and the many-task setting. Despite its simplicity, the algorithm matches, up to a constant factor, a lower bound on the performance of any such parameter-transfer method under natural task similarity assumptions. We use experiments in both convex and deep learning settings to verify and demonstrate the applicability of our theory.'
volume: 97
URL: https://proceedings.mlr.press/v97/balcan19a.html
PDF: http://proceedings.mlr.press/v97/balcan19a/balcan19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-balcan19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Maria-Florina
family: Balcan
- given: Mikhail
family: Khodak
- given: Ameet
family: Talwalkar
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 424-433
id: balcan19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 424
lastpage: 433
published: 2019-05-24 00:00:00 +0000
- title: 'Open-ended learning in symmetric zero-sum games'
abstract: 'Zero-sum games such as chess and poker are, abstractly, functions that evaluate pairs of agents, for example labeling them ‘winner’ and ‘loser’. If the game is approximately transitive, then self-play generates sequences of agents of increasing strength. However, nontransitive games, such as rock-paper-scissors, can exhibit strategic cycles, and there is no longer a clear objective – we want agents to increase in strength, but against whom is unclear. In this paper, we introduce a geometric framework for formulating agent objectives in zero-sum games, in order to construct adaptive sequences of objectives that yield open-ended learning. The framework allows us to reason about population performance in nontransitive games, and enables the development of a new algorithm (rectified Nash response, PSRO_rN) that uses game-theoretic niching to construct diverse populations of effective agents, producing a stronger set of agents than existing algorithms. We apply PSRO_rN to two highly nontransitive resource allocation games and find that PSRO_rN consistently outperforms the existing alternatives.'
volume: 97
URL: https://proceedings.mlr.press/v97/balduzzi19a.html
PDF: http://proceedings.mlr.press/v97/balduzzi19a/balduzzi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-balduzzi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: David
family: Balduzzi
- given: Marta
family: Garnelo
- given: Yoram
family: Bachrach
- given: Wojciech
family: Czarnecki
- given: Julien
family: Perolat
- given: Max
family: Jaderberg
- given: Thore
family: Graepel
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 434-443
id: balduzzi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 434
lastpage: 443
published: 2019-05-24 00:00:00 +0000
- title: 'Concrete Autoencoders: Differentiable Feature Selection and Reconstruction'
abstract: 'We introduce the concrete autoencoder, an end-to-end differentiable method for global feature selection, which efficiently identifies a subset of the most informative features and simultaneously learns a neural network to reconstruct the input data from the selected features. Our method is unsupervised, and is based on using a concrete selector layer as the encoder and using a standard neural network as the decoder. During the training phase, the temperature of the concrete selector layer is gradually decreased, which encourages a user-specified number of discrete features to be learned; during test time, the selected features can be used with the decoder network to reconstruct the remaining input features. We evaluate concrete autoencoders on a variety of datasets, where they significantly outperform state-of-the-art methods for feature selection and data reconstruction. In particular, on a large-scale gene expression dataset, the concrete autoencoder selects a small subset of genes whose expression levels can be used to impute the expression levels of the remaining genes; in doing so, it improves on the current widely-used expert-curated L1000 landmark genes, potentially reducing measurement costs by 20%. The concrete autoencoder can be implemented by adding just a few lines of code to a standard autoencoder, and the code for the algorithm and experiments is publicly available.'
volume: 97
URL: https://proceedings.mlr.press/v97/balin19a.html
PDF: http://proceedings.mlr.press/v97/balin19a/balin19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-balin19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Muhammed Fatih
family: Balın
- given: Abubakar
family: Abid
- given: James
family: Zou
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 444-453
id: balin19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 444
lastpage: 453
published: 2019-05-24 00:00:00 +0000
- title: 'HOList: An Environment for Machine Learning of Higher Order Logic Theorem Proving'
abstract: 'We present an environment, benchmark, and deep learning driven automated theorem prover for higher-order logic. Higher-order interactive theorem provers enable the formalization of arbitrary mathematical theories and thereby present an interesting challenge for deep learning. We provide an open-source framework based on the HOL Light theorem prover that can be used as a reinforcement learning environment. HOL Light comes with a broad coverage of basic mathematical theorems on calculus and the formal proof of the Kepler conjecture, from which we derive a challenging benchmark for automated reasoning approaches. We also present a deep reinforcement learning driven automated theorem prover, DeepHOL, that gives strong initial results on this benchmark.'
volume: 97
URL: https://proceedings.mlr.press/v97/bansal19a.html
PDF: http://proceedings.mlr.press/v97/bansal19a/bansal19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-bansal19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Kshitij
family: Bansal
- given: Sarah
family: Loos
- given: Markus
family: Rabe
- given: Christian
family: Szegedy
- given: Stewart
family: Wilcox
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 454-463
id: bansal19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 454
lastpage: 463
published: 2019-05-24 00:00:00 +0000
- title: 'Structured agents for physical construction'
abstract: 'Physical construction—the ability to compose objects, subject to physical dynamics, to serve some function—is fundamental to human intelligence. We introduce a suite of challenging physical construction tasks inspired by how children play with blocks, such as matching a target configuration, stacking blocks to connect objects together, and creating shelter-like structures over target objects. We examine how a range of deep reinforcement learning agents fare on these challenges, and introduce several new approaches which provide superior performance. Our results show that agents which use structured representations (e.g., objects and scene graphs) and structured policies (e.g., object-centric actions) outperform those which use less structured representations, and generalize better beyond their training when asked to reason about larger scenes. Model-based agents which use Monte-Carlo Tree Search also outperform strictly model-free agents in our most challenging construction problems. We conclude that approaches which combine structured representations and reasoning with powerful learning are a key path toward agents that possess rich intuitive physics, scene understanding, and planning.'
volume: 97
URL: https://proceedings.mlr.press/v97/bapst19a.html
PDF: http://proceedings.mlr.press/v97/bapst19a/bapst19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-bapst19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Victor
family: Bapst
- given: Alvaro
family: Sanchez-Gonzalez
- given: Carl
family: Doersch
- given: Kimberly
family: Stachenfeld
- given: Pushmeet
family: Kohli
- given: Peter
family: Battaglia
- given: Jessica
family: Hamrick
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 464-474
id: bapst19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 464
lastpage: 474
published: 2019-05-24 00:00:00 +0000
- title: 'Learning to Route in Similarity Graphs'
abstract: 'Recently similarity graphs became the leading paradigm for efficient nearest neighbor search, outperforming traditional tree-based and LSH-based methods. Similarity graphs perform the search via greedy routing: a query traverses the graph and in each vertex moves to the adjacent vertex that is the closest to this query. In practice, similarity graphs are often susceptible to local minima, when queries do not reach its nearest neighbors, getting stuck in suboptimal vertices. In this paper we propose to learn the routing function that overcomes local minima via incorporating information about the graph global structure. In particular, we augment the vertices of a given graph with additional representations that are learned to provide the optimal routing from the start vertex to the query nearest neighbor. By thorough experiments, we demonstrate that the proposed learnable routing successfully diminishes the local minima problem and significantly improves the overall search performance.'
volume: 97
URL: https://proceedings.mlr.press/v97/baranchuk19a.html
PDF: http://proceedings.mlr.press/v97/baranchuk19a/baranchuk19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-baranchuk19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Dmitry
family: Baranchuk
- given: Dmitry
family: Persiyanov
- given: Anton
family: Sinitsin
- given: Artem
family: Babenko
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 475-484
id: baranchuk19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 475
lastpage: 484
published: 2019-05-24 00:00:00 +0000
- title: 'A Personalized Affective Memory Model for Improving Emotion Recognition'
abstract: 'Recent models of emotion recognition strongly rely on supervised deep learning solutions for the distinction of general emotion expressions. However, they are not reliable when recognizing online and personalized facial expressions, e.g., for person-specific affective understanding. In this paper, we present a neural model based on a conditional adversarial autoencoder to learn how to represent and edit general emotion expressions. We then propose Grow-When-Required networks as personalized affective memories to learn individualized aspects of emotional expressions. Our model achieves state-of-the-art performance on emotion recognition when evaluated on in-the-wild datasets. Furthermore, our experiments include ablation studies and neural visualizations in order to explain the behavior of our model.'
volume: 97
URL: https://proceedings.mlr.press/v97/barros19a.html
PDF: http://proceedings.mlr.press/v97/barros19a/barros19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-barros19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Pablo
family: Barros
- given: German
family: Parisi
- given: Stefan
family: Wermter
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 485-494
id: barros19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 485
lastpage: 494
published: 2019-05-24 00:00:00 +0000
- title: 'Scale-free adaptive planning for deterministic dynamics & discounted rewards'
abstract: 'We address the problem of planning in an environment with deterministic dynamics and stochastic discounted rewards under a limited numerical budget where the ranges of both rewards and noise are unknown. We introduce PlaTypOOS, an adaptive, robust, and efficient alternative to the OLOP (open-loop optimistic planning) algorithm. Whereas OLOP requires a priori knowledge of the ranges of both rewards and noise, PlaTypOOS dynamically adapts its behavior to both. This allows PlaTypOOS to be immune to two vulnerabilities of OLOP: failure when given underestimated ranges of noise and rewards and inefficiency when these are overestimated. PlaTypOOS additionally adapts to the global smoothness of the value function. PlaTypOOS acts in a provably more efficient manner vs. OLOP when OLOP is given an overestimated reward and show that in the case of no noise, PlaTypOOS learns exponentially faster.'
volume: 97
URL: https://proceedings.mlr.press/v97/bartlett19a.html
PDF: http://proceedings.mlr.press/v97/bartlett19a/bartlett19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-bartlett19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Peter
family: Bartlett
- given: Victor
family: Gabillon
- given: Jennifer
family: Healey
- given: Michal
family: Valko
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 495-504
id: bartlett19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 495
lastpage: 504
published: 2019-05-24 00:00:00 +0000
- title: 'Pareto Optimal Streaming Unsupervised Classification'
abstract: 'We study an online and streaming unsupervised classification system. Our setting consists of a collection of classifiers (with unknown confusion matrices) each of which can classify one sample per unit time, and which are accessed by a stream of unlabeled samples. Each sample is dispatched to one or more classifiers, and depending on the labels collected from these classifiers, may be sent to other classifiers to collect additional labels. The labels are continually aggregated. Once the aggregated label has high enough accuracy (a pre-specified threshold for accuracy) or the sample is sent to all the classifiers, the now labeled sample is ejected from the system. For any given pre-specified threshold for accuracy, the objective is to sustain the maximum possible rate of arrival of new samples, such that the number of samples in memory does not grow unbounded. In this paper, we characterize the Pareto-optimal region of accuracy and arrival rate, and develop an algorithm that can operate at any point within this region. Our algorithm uses queueing-based routing and scheduling approaches combined with novel online tensor decomposition method to learn the hidden parameters, to Pareto-optimality guarantees. We finally verify our theoretical results through simulations on two ensembles formed using AlexNet, VGG, and ResNet deep image classifiers.'
volume: 97
URL: https://proceedings.mlr.press/v97/basu19a.html
PDF: http://proceedings.mlr.press/v97/basu19a/basu19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-basu19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Soumya
family: Basu
- given: Steven
family: Gutstein
- given: Brent
family: Lance
- given: Sanjay
family: Shakkottai
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 505-514
id: basu19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 505
lastpage: 514
published: 2019-05-24 00:00:00 +0000
- title: 'Categorical Feature Compression via Submodular Optimization'
abstract: 'In the era of big data, learning from categorical features with very large vocabularies (e.g., 28 million for the Criteo click prediction dataset) has become a practical challenge for machine learning researchers and practitioners. We design a highly-scalable vocabulary compression algorithm that seeks to maximize the mutual information between the compressed categorical feature and the target binary labels and we furthermore show that its solution is guaranteed to be within a $1-1/e \approx 63%$ factor of the global optimal solution. Although in some settings, entropy-based set functions are known to be submodular, this is not the case for the mutual information objective we consider (mutual information with respect to the target labels). To address this, we introduce a novel re-parametrization of the mutual information objective, which we prove is submodular, and also design a data structure to query the submodular function in amortized $O(\log n )$ time (where $n$ is the input vocabulary size). Our complete algorithm is shown to operate in $O(n \log n )$ time. Additionally, we design a distributed implementation in which the query data structure is decomposed across $O(k)$ machines such that each machine only requires $O(\frac n k)$ space, while still preserving the approximation guarantee and using only logarithmic rounds of computation. We also provide analysis of simple alternative heuristic compression methods to demonstrate they cannot achieve any approximation guarantee. Using the large-scale Criteo learning task, we demonstrate better performance in retaining mutual information and also verify competitive learning performance compared to other baseline methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/bateni19a.html
PDF: http://proceedings.mlr.press/v97/bateni19a/bateni19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-bateni19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Mohammadhossein
family: Bateni
- given: Lin
family: Chen
- given: Hossein
family: Esfandiari
- given: Thomas
family: Fu
- given: Vahab
family: Mirrokni
- given: Afshin
family: Rostamizadeh
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 515-523
id: bateni19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 515
lastpage: 523
published: 2019-05-24 00:00:00 +0000
- title: 'Noise2Self: Blind Denoising by Self-Supervision'
abstract: 'We propose a general framework for denoising high-dimensional measurements which requires no prior on the signal, no estimate of the noise, and no clean training data. The only assumption is that the noise exhibits statistical independence across different dimensions of the measurement, while the true signal exhibits some correlation. For a broad class of functions (“$\mathcal{J}$-invariant”), it is then possible to estimate the performance of a denoiser from noisy data alone. This allows us to calibrate $\mathcal{J}$-invariant versions of any parameterised denoising algorithm, from the single hyperparameter of a median filter to the millions of weights of a deep neural network. We demonstrate this on natural image and microscopy data, where we exploit noise independence between pixels, and on single-cell gene expression data, where we exploit independence between detections of individual molecules. This framework generalizes recent work on training neural nets from noisy images and on cross-validation for matrix factorization.'
volume: 97
URL: https://proceedings.mlr.press/v97/batson19a.html
PDF: http://proceedings.mlr.press/v97/batson19a/batson19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-batson19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Joshua
family: Batson
- given: Loic
family: Royer
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 524-533
id: batson19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 524
lastpage: 533
published: 2019-05-24 00:00:00 +0000
- title: 'Efficient optimization of loops and limits with randomized telescoping sums'
abstract: 'We consider optimization problems in which the objective requires an inner loop with many steps or is the limit of a sequence of increasingly costly approximations. Meta-learning, training recurrent neural networks, and optimization of the solutions to differential equations are all examples of optimization problems with this character. In such problems, it can be expensive to compute the objective function value and its gradient, but truncating the loop or using less accurate approximations can induce biases that damage the overall solution. We propose *randomized telescope* (RT) gradient estimators, which represent the objective as the sum of a telescoping series and sample linear combinations of terms to provide cheap unbiased gradient estimates. We identify conditions under which RT estimators achieve optimization convergence rates independent of the length of the loop or the required accuracy of the approximation. We also derive a method for tuning RT estimators online to maximize a lower bound on the expected decrease in loss per unit of computation. We evaluate our adaptive RT estimators on a range of applications including meta-optimization of learning rates, variational inference of ODE parameters, and training an LSTM to model long sequences.'
volume: 97
URL: https://proceedings.mlr.press/v97/beatson19a.html
PDF: http://proceedings.mlr.press/v97/beatson19a/beatson19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-beatson19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alex
family: Beatson
- given: Ryan P
family: Adams
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 534-543
id: beatson19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 534
lastpage: 543
published: 2019-05-24 00:00:00 +0000
- title: 'Recurrent Kalman Networks: Factorized Inference in High-Dimensional Deep Feature Spaces'
abstract: 'In order to integrate uncertainty estimates into deep time-series modelling, Kalman Filters (KFs) (Kalman et al., 1960) have been integrated with deep learning models, however, such approaches typically rely on approximate inference tech- niques such as variational inference which makes learning more complex and often less scalable due to approximation errors. We propose a new deep approach to Kalman filtering which can be learned directly in an end-to-end manner using backpropagation without additional approximations. Our approach uses a high-dimensional factorized latent state representation for which the Kalman updates simplify to scalar operations and thus avoids hard to backpropagate, computationally heavy and potentially unstable matrix inversions. Moreover, we use locally linear dynamic models to efficiently propagate the latent state to the next time step. The resulting network architecture, which we call Recurrent Kalman Network (RKN), can be used for any time-series data, similar to a LSTM (Hochreiter & Schmidhuber, 1997) but uses an explicit representation of uncertainty. As shown by our experiments, the RKN obtains much more accurate uncertainty estimates than an LSTM or Gated Recurrent Units (GRUs) (Cho et al., 2014) while also showing a slightly improved prediction performance and outperforms various recent generative models on an image imputation task.'
volume: 97
URL: https://proceedings.mlr.press/v97/becker19a.html
PDF: http://proceedings.mlr.press/v97/becker19a/becker19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-becker19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Philipp
family: Becker
- given: Harit
family: Pandya
- given: Gregor
family: Gebhardt
- given: Cheng
family: Zhao
- given: C. James
family: Taylor
- given: Gerhard
family: Neumann
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 544-552
id: becker19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 544
lastpage: 552
published: 2019-05-24 00:00:00 +0000
- title: 'Switching Linear Dynamics for Variational Bayes Filtering'
abstract: 'System identification of complex and nonlinear systems is a central problem for model predictive control and model-based reinforcement learning. Despite their complexity, such systems can often be approximated well by a set of linear dynamical systems if broken into appropriate subsequences. This mechanism not only helps us find good approximations of dynamics, but also gives us deeper insight into the underlying system. Leveraging Bayesian inference, Variational Autoencoders and Concrete relaxations, we show how to learn a richer and more meaningful state space, e.g. encoding joint constraints and collisions with walls in a maze, from partial and high-dimensional observations. This representation translates into a gain of accuracy of learned dynamics showcased on various simulated tasks.'
volume: 97
URL: https://proceedings.mlr.press/v97/becker-ehmck19a.html
PDF: http://proceedings.mlr.press/v97/becker-ehmck19a/becker-ehmck19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-becker-ehmck19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Philip
family: Becker-Ehmck
- given: Jan
family: Peters
- given: Patrick
family: Van Der Smagt
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 553-562
id: becker-ehmck19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 553
lastpage: 562
published: 2019-05-24 00:00:00 +0000
- title: 'Active Learning for Probabilistic Structured Prediction of Cuts and Matchings'
abstract: 'Active learning methods, like uncertainty sampling, combined with probabilistic prediction techniques have achieved success in various problems like image classification and text classification. For more complex multivariate prediction tasks, the relationships between labels play an important role in designing structured classifiers with better performance. However, computational time complexity limits prevalent probabilistic methods from effectively supporting active learning. Specifically, while non-probabilistic methods based on structured support vector ma-chines can be tractably applied to predicting cuts and bipartite matchings, conditional random fields are intractable for these structures. We propose an adversarial approach for active learning with structured prediction domains that is tractable for cuts and matching. We evaluate this approach algorithmically in two important structured prediction problems: multi-label classification and object tracking in videos. We demonstrate better accuracy and computational efficiency for our proposed method.'
volume: 97
URL: https://proceedings.mlr.press/v97/behpour19a.html
PDF: http://proceedings.mlr.press/v97/behpour19a/behpour19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-behpour19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sima
family: Behpour
- given: Anqi
family: Liu
- given: Brian
family: Ziebart
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 563-572
id: behpour19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 563
lastpage: 572
published: 2019-05-24 00:00:00 +0000
- title: 'Invertible Residual Networks'
abstract: 'We show that standard ResNet architectures can be made invertible, allowing the same model to be used for classification, density estimation, and generation. Typically, enforcing invertibility requires partitioning dimensions or restricting network architectures. In contrast, our approach only requires adding a simple normalization step during training, already available in standard frameworks. Invertible ResNets define a generative model which can be trained by maximum likelihood on unlabeled data. To compute likelihoods, we introduce a tractable approximation to the Jacobian log-determinant of a residual block. Our empirical evaluation shows that invertible ResNets perform competitively with both state-of-the-art image classifiers and flow-based generative models, something that has not been previously achieved with a single architecture.'
volume: 97
URL: https://proceedings.mlr.press/v97/behrmann19a.html
PDF: http://proceedings.mlr.press/v97/behrmann19a/behrmann19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-behrmann19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jens
family: Behrmann
- given: Will
family: Grathwohl
- given: Ricky T. Q.
family: Chen
- given: David
family: Duvenaud
- given: Joern-Henrik
family: Jacobsen
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 573-582
id: behrmann19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 573
lastpage: 582
published: 2019-05-24 00:00:00 +0000
- title: 'Greedy Layerwise Learning Can Scale To ImageNet'
abstract: 'Shallow supervised 1-hidden layer neural networks have a number of favorable properties that make them easier to interpret, analyze, and optimize than their deep counterparts, but lack their representational power. Here we use 1-hidden layer learning problems to sequentially build deep networks layer by layer, which can inherit properties from shallow networks. Contrary to previous approaches using shallow networks, we focus on problems where deep learning is reported as critical for success. We thus study CNNs on image classification tasks using the large-scale ImageNet dataset and the CIFAR-10 dataset. Using a simple set of ideas for architecture and training we find that solving sequential 1-hidden-layer auxiliary problems lead to a CNN that exceeds AlexNet performance on ImageNet. Extending this training methodology to construct individual layers by solving 2-and-3-hidden layer auxiliary problems, we obtain an 11-layer network that exceeds several members of the VGG model family on ImageNet, and can train a VGG-11 model to the same accuracy as end-to-end learning. To our knowledge, this is the first competitive alternative to end-to-end training of CNNs that can scale to ImageNet. We illustrate several interesting properties of these models and conduct a range of experiments to study the properties this training induces on the intermediate layers.'
volume: 97
URL: https://proceedings.mlr.press/v97/belilovsky19a.html
PDF: http://proceedings.mlr.press/v97/belilovsky19a/belilovsky19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-belilovsky19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Eugene
family: Belilovsky
- given: Michael
family: Eickenberg
- given: Edouard
family: Oyallon
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 583-593
id: belilovsky19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 583
lastpage: 593
published: 2019-05-24 00:00:00 +0000
- title: 'Overcoming Multi-model Forgetting'
abstract: 'We identify a phenomenon, which we refer to as multi-model forgetting, that occurs when sequentially training multiple deep networks with partially-shared parameters; the performance of previously-trained models degrades as one optimizes a subsequent one, due to the overwriting of shared parameters. To overcome this, we introduce a statistically-justified weight plasticity loss that regularizes the learning of a model’s shared parameters according to their importance for the previous models, and demonstrate its effectiveness when training two models sequentially and for neural architecture search. Adding weight plasticity in neural architecture search preserves the best models to the end of the search and yields improved results in both natural language processing and computer vision tasks.'
volume: 97
URL: https://proceedings.mlr.press/v97/benyahia19a.html
PDF: http://proceedings.mlr.press/v97/benyahia19a/benyahia19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-benyahia19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yassine
family: Benyahia
- given: Kaicheng
family: Yu
- given: Kamil Bennani
family: Smires
- given: Martin
family: Jaggi
- given: Anthony C.
family: Davison
- given: Mathieu
family: Salzmann
- given: Claudiu
family: Musat
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 594-603
id: benyahia19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 594
lastpage: 603
published: 2019-05-24 00:00:00 +0000
- title: 'Optimal Kronecker-Sum Approximation of Real Time Recurrent Learning'
abstract: 'One of the central goals of Recurrent Neural Networks (RNNs) is to learn long-term dependencies in sequential data. Nevertheless, the most popular training method, Truncated Backpropagation through Time (TBPTT), categorically forbids learning dependencies beyond the truncation horizon. In contrast, the online training algorithm Real Time Recurrent Learning (RTRL) provides untruncated gradients, with the disadvantage of impractically large computational costs. Recently published approaches reduce these costs by providing noisy approximations of RTRL. We present a new approximation algorithm of RTRL, Optimal Kronecker-Sum Approximation (OK). We prove that OK is optimal for a class of approximations of RTRL, which includes all approaches published so far. Additionally, we show that OK has empirically negligible noise: Unlike previous algorithms it matches TBPTT in a real world task (character-level Penn TreeBank) and can exploit online parameter updates to outperform TBPTT in a synthetic string memorization task. Code available at GitHub.'
volume: 97
URL: https://proceedings.mlr.press/v97/benzing19a.html
PDF: http://proceedings.mlr.press/v97/benzing19a/benzing19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-benzing19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Frederik
family: Benzing
- given: Marcelo Matheus
family: Gauy
- given: Asier
family: Mujika
- given: Anders
family: Martinsson
- given: Angelika
family: Steger
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 604-613
id: benzing19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 604
lastpage: 613
published: 2019-05-24 00:00:00 +0000
- title: 'Adversarially Learned Representations for Information Obfuscation and Inference'
abstract: 'Data collection and sharing are pervasive aspects of modern society. This process can either be voluntary, as in the case of a person taking a facial image to unlock his/her phone, or incidental, such as traffic cameras collecting videos on pedestrians. An undesirable side effect of these processes is that shared data can carry information about attributes that users might consider as sensitive, even when such information is of limited use for the task. It is therefore desirable for both data collectors and users to design procedures that minimize sensitive information leakage. Balancing the competing objectives of providing meaningful individualized service levels and inference while obfuscating sensitive information is still an open problem. In this work, we take an information theoretic approach that is implemented as an unconstrained adversarial game between Deep Neural Networks in a principled, data-driven manner. This approach enables us to learn domain-preserving stochastic transformations that maintain performance on existing algorithms while minimizing sensitive information leakage.'
volume: 97
URL: https://proceedings.mlr.press/v97/bertran19a.html
PDF: http://proceedings.mlr.press/v97/bertran19a/bertran19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-bertran19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Martin
family: Bertran
- given: Natalia
family: Martinez
- given: Afroditi
family: Papadaki
- given: Qiang
family: Qiu
- given: Miguel
family: Rodrigues
- given: Galen
family: Reeves
- given: Guillermo
family: Sapiro
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 614-623
id: bertran19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 614
lastpage: 623
published: 2019-05-24 00:00:00 +0000
- title: 'Bandit Multiclass Linear Classification: Efficient Algorithms for the Separable Case'
abstract: 'We study the problem of efficient online multiclass linear classification with bandit feedback, where all examples belong to one of $K$ classes and lie in the $d$-dimensional Euclidean space. Previous works have left open the challenge of designing efficient algorithms with finite mistake bounds when the data is linearly separable by a margin $\gamma$. In this work, we take a first step towards this problem. We consider two notions of linear separability: strong and weak. 1. Under the strong linear separability condition, we design an efficient algorithm that achieves a near-optimal mistake bound of $O\left(\frac{K}{\gamma^2} \right)$. 2. Under the more challenging weak linear separability condition, we design an efficient algorithm with a mistake bound of $2^{\widetilde{O}(\min(K \log^2 \frac{1}{\gamma}, \sqrt{\frac{1}{\gamma}} \log K))}$. Our algorithm is based on kernel Perceptron, which is inspired by the work of Klivans & Servedio (2008) on improperly learning intersection of halfspaces.'
volume: 97
URL: https://proceedings.mlr.press/v97/beygelzimer19a.html
PDF: http://proceedings.mlr.press/v97/beygelzimer19a/beygelzimer19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-beygelzimer19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alina
family: Beygelzimer
- given: David
family: Pal
- given: Balazs
family: Szorenyi
- given: Devanathan
family: Thiruvenkatachari
- given: Chen-Yu
family: Wei
- given: Chicheng
family: Zhang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 624-633
id: beygelzimer19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 624
lastpage: 633
published: 2019-05-24 00:00:00 +0000
- title: 'Analyzing Federated Learning through an Adversarial Lens'
abstract: 'Federated learning distributes model training among a multitude of agents, who, guided by privacy concerns, perform training using their local data but share only model parameter updates, for iterative aggregation at the server to train an overall global model. In this work, we explore how the federated learning setting gives rise to a new threat, namely model poisoning, which differs from traditional data poisoning. Model poisoning is carried out by an adversary controlling a small number of malicious agents (usually 1) with the aim of causing the global model to misclassify a set of chosen inputs with high conﬁdence. We explore a number of strategies to carry out this attack on deep neural networks, starting with targeted model poisoning using a simple boosting of the malicious agent’s update to overcome the effects of other agents. We also propose two critical notions of stealth to detect malicious updates. We bypass these by including them in the adversarial objective to carry out stealthy model poisoning. We improve its stealth with the use of an alternating minimization strategy which alternately optimizes for stealth and the adversarial objective. We also empirically demonstrate that Byzantine-resilient aggregation strategies are not robust to our attacks. Our results indicate that highly constrained adversaries can carry out model poisoning attacks while maintaining stealth, thus highlighting the vulnerability of the federated learning setting and the need to develop effective defense strategies.'
volume: 97
URL: https://proceedings.mlr.press/v97/bhagoji19a.html
PDF: http://proceedings.mlr.press/v97/bhagoji19a/bhagoji19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-bhagoji19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Arjun Nitin
family: Bhagoji
- given: Supriyo
family: Chakraborty
- given: Prateek
family: Mittal
- given: Seraphin
family: Calo
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 634-643
id: bhagoji19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 634
lastpage: 643
published: 2019-05-24 00:00:00 +0000
- title: 'Optimal Continuous DR-Submodular Maximization and Applications to Provable Mean Field Inference'
abstract: 'Mean field inference for discrete graphical models is generally a highly nonconvex problem, which also holds for the class of probabilistic log-submodular models. Existing optimization methods, e.g., coordinate ascent algorithms, typically only find local optima. In this work we propose provable mean filed methods for probabilistic log-submodular models and its posterior agreement (PA) with strong approximation guarantees. The main algorithmic technique is a new Double Greedy scheme, termed DR-DoubleGreedy, for continuous DR-submodular maximization with box-constraints. It is a one-pass algorithm with linear time complexity, reaching the optimal 1/2 approximation ratio, which may be of independent interest. We validate the superior performance of our algorithms against baselines on both synthetic and real-world datasets.'
volume: 97
URL: https://proceedings.mlr.press/v97/bian19a.html
PDF: http://proceedings.mlr.press/v97/bian19a/bian19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-bian19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yatao
family: Bian
- given: Joachim
family: Buhmann
- given: Andreas
family: Krause
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 644-653
id: bian19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 644
lastpage: 653
published: 2019-05-24 00:00:00 +0000
- title: 'More Efficient Off-Policy Evaluation through Regularized Targeted Learning'
abstract: 'We study the problem of off-policy evaluation (OPE) in Reinforcement Learning (RL), where the aim is to estimate the performance of a new policy given historical data that may have been generated by a different policy, or policies. In particular, we introduce a novel doubly-robust estimator for the OPE problem in RL, based on the Targeted Maximum Likelihood Estimation principle from the statistical causal inference literature. We also introduce several variance reduction techniques that lead to impressive performance gains in off-policy evaluation. We show empirically that our estimator uniformly wins over existing off-policy evaluation methods across multiple RL environments and various levels of model misspecification. Finally, we further the existing theoretical analysis of estimators for the RL off-policy estimation problem by showing their $O_P(1/\sqrt{n})$ rate of convergence and characterizing their asymptotic distribution.'
volume: 97
URL: https://proceedings.mlr.press/v97/bibaut19a.html
PDF: http://proceedings.mlr.press/v97/bibaut19a/bibaut19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-bibaut19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Aurelien
family: Bibaut
- given: Ivana
family: Malenica
- given: Nikos
family: Vlassis
- given: Mark
family: Van Der Laan
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 654-663
id: bibaut19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 654
lastpage: 663
published: 2019-05-24 00:00:00 +0000
- title: 'A Kernel Perspective for Regularizing Deep Neural Networks'
abstract: 'We propose a new point of view for regularizing deep neural networks by using the norm of a reproducing kernel Hilbert space (RKHS). Even though this norm cannot be computed, it admits upper and lower approximations leading to various practical strategies. Specifically, this perspective (i) provides a common umbrella for many existing regularization principles, including spectral norm and gradient penalties, or adversarial training, (ii) leads to new effective regularization penalties, and (iii) suggests hybrid strategies combining lower and upper bounds to get better approximations of the RKHS norm. We experimentally show this approach to be effective when learning on small datasets, or to obtain adversarially robust models.'
volume: 97
URL: https://proceedings.mlr.press/v97/bietti19a.html
PDF: http://proceedings.mlr.press/v97/bietti19a/bietti19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-bietti19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alberto
family: Bietti
- given: Grégoire
family: Mialon
- given: Dexiong
family: Chen
- given: Julien
family: Mairal
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 664-674
id: bietti19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 664
lastpage: 674
published: 2019-05-24 00:00:00 +0000
- title: 'Rethinking Lossy Compression: The Rate-Distortion-Perception Tradeoff'
abstract: 'Lossy compression algorithms are typically designed and analyzed through the lens of Shannon’s rate-distortion theory, where the goal is to achieve the lowest possible distortion (e.g., low MSE or high SSIM) at any given bit rate. However, in recent years, it has become increasingly accepted that "low distortion" is not a synonym for "high perceptual quality", and in fact optimization of one often comes at the expense of the other. In light of this understanding, it is natural to seek for a generalization of rate-distortion theory which takes perceptual quality into account. In this paper, we adopt the mathematical definition of perceptual quality recently proposed by Blau & Michaeli (2018), and use it to study the three-way tradeoff between rate, distortion, and perception. We show that restricting the perceptual quality to be high, generally leads to an elevation of the rate-distortion curve, thus necessitating a sacrifice in either rate or distortion. We prove several fundamental properties of this triple-tradeoff, calculate it in closed form for a Bernoulli source, and illustrate it visually on a toy MNIST example.'
volume: 97
URL: https://proceedings.mlr.press/v97/blau19a.html
PDF: http://proceedings.mlr.press/v97/blau19a/blau19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-blau19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yochai
family: Blau
- given: Tomer
family: Michaeli
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 675-685
id: blau19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 675
lastpage: 685
published: 2019-05-24 00:00:00 +0000
- title: 'Correlated bandits or: How to minimize mean-squared error online'
abstract: 'While the objective in traditional multi-armed bandit problems is to find the arm with the highest mean, in many settings, finding an arm that best captures information about other arms is of interest. This objective, however, requires learning the underlying correlation structure and not just the means. Sensors placement for industrial surveillance and cellular network monitoring are a few applications, where the underlying correlation structure plays an important role. Motivated by such applications, we formulate the correlated bandit problem, where the objective is to find the arm with the lowest mean-squared error (MSE) in estimating all the arms. To this end, we derive first an MSE estimator based on sample variances/covariances and show that our estimator exponentially concentrates around the true MSE. Under a best-arm identification framework, we propose a successive rejects type algorithm and provide bounds on the probability of error in identifying the best arm. Using minimax theory, we also derive fundamental performance limits for the correlated bandit problem.'
volume: 97
URL: https://proceedings.mlr.press/v97/boda19a.html
PDF: http://proceedings.mlr.press/v97/boda19a/boda19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-boda19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Vinay Praneeth
family: Boda
- given: Prashanth
family: L.A.
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 686-694
id: boda19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 686
lastpage: 694
published: 2019-05-24 00:00:00 +0000
- title: 'Adversarial Attacks on Node Embeddings via Graph Poisoning'
abstract: 'The goal of network representation learning is to learn low-dimensional node embeddings that capture the graph structure and are useful for solving downstream tasks. However, despite the proliferation of such methods, there is currently no study of their robustness to adversarial attacks. We provide the first adversarial vulnerability analysis on the widely used family of methods based on random walks. We derive efficient adversarial perturbations that poison the network structure and have a negative effect on both the quality of the embeddings and the downstream tasks. We further show that our attacks are transferable since they generalize to many models and are successful even when the attacker is restricted.'
volume: 97
URL: https://proceedings.mlr.press/v97/bojchevski19a.html
PDF: http://proceedings.mlr.press/v97/bojchevski19a/bojchevski19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-bojchevski19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Aleksandar
family: Bojchevski
- given: Stephan
family: Günnemann
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 695-704
id: bojchevski19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 695
lastpage: 704
published: 2019-05-24 00:00:00 +0000
- title: 'Online Variance Reduction with Mixtures'
abstract: 'Adaptive importance sampling for stochastic optimization is a promising approach that offers improved convergence through variance reduction. In this work, we propose a new framework for variance reduction that enables the use of mixtures over predefined sampling distributions, which can naturally encode prior knowledge about the data. While these sampling distributions are fixed, the mixture weights are adapted during the optimization process. We propose VRM, a novel and efficient adaptive scheme that asymptotically recovers the best mixture weights in hindsight and can also accommodate sampling distributions over sets of points. We empirically demonstrate the versatility of VRM in a range of applications.'
volume: 97
URL: https://proceedings.mlr.press/v97/borsos19a.html
PDF: http://proceedings.mlr.press/v97/borsos19a/borsos19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-borsos19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Zalán
family: Borsos
- given: Sebastian
family: Curi
- given: Kfir Yehuda
family: Levy
- given: Andreas
family: Krause
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 705-714
id: borsos19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 705
lastpage: 714
published: 2019-05-24 00:00:00 +0000
- title: 'Compositional Fairness Constraints for Graph Embeddings'
abstract: 'Learning high-quality node embeddings is a key building block for machine learning models that operate on graph data, such as social networks and recommender systems. However, existing graph embedding techniques are unable to cope with fairness constraints, e.g., ensuring that the learned representations do not correlate with certain attributes, such as age or gender. Here, we introduce an adversarial framework to enforce fairness constraints on graph embeddings. Our approach is *compositional*—meaning that it can flexibly accommodate different combinations of fairness constraints during inference. For instance, in the context of social recommendations, our framework would allow one user to request that their recommendations are invariant to both their age and gender, while also allowing another user to request invariance to just their age. Experiments on standard knowledge graph and recommender system benchmarks highlight the utility of our proposed framework.'
volume: 97
URL: https://proceedings.mlr.press/v97/bose19a.html
PDF: http://proceedings.mlr.press/v97/bose19a/bose19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-bose19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Avishek
family: Bose
- given: William
family: Hamilton
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 715-724
id: bose19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 715
lastpage: 724
published: 2019-05-24 00:00:00 +0000
- title: 'Unreproducible Research is Reproducible'
abstract: 'The apparent contradiction in the title is a wordplay on the different meanings attributed to the word reproducible across different scientific fields. What we imply is that unreproducible findings can be built upon reproducible methods. Without denying the importance of facilitating the reproduction of methods, we deem important to reassert that reproduction of findings is a fundamental step of the scientific inquiry. We argue that the commendable quest towards easy deterministic reproducibility of methods and numerical results should not have us forget the even more important necessity of ensuring the reproducibility of empirical findings and conclusions by properly accounting for essential sources of variations. We provide experiments to exemplify the brittleness of current common practice in the evaluation of models in the field of deep learning, showing that even if the results could be reproduced, a slightly different experiment would not support the findings. We hope to help clarify the distinction between exploratory and empirical research in the field of deep learning and believe more energy should be devoted to proper empirical research in our community. This work is an attempt to promote the use of more rigorous and diversified methodologies. It is not an attempt to impose a new methodology and it is not a critique on the nature of exploratory research.'
volume: 97
URL: https://proceedings.mlr.press/v97/bouthillier19a.html
PDF: http://proceedings.mlr.press/v97/bouthillier19a/bouthillier19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-bouthillier19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Xavier
family: Bouthillier
- given: César
family: Laurent
- given: Pascal
family: Vincent
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 725-734
id: bouthillier19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 725
lastpage: 734
published: 2019-05-24 00:00:00 +0000
- title: 'Blended Conditonal Gradients'
abstract: 'We present a blended conditional gradient approach for minimizing a smooth convex function over a polytope P, combining the Frank{–}Wolfe algorithm (also called conditional gradient) with gradient-based steps, different from away steps and pairwise steps, but still achieving linear convergence for strongly convex functions, along with good practical performance. Our approach retains all favorable properties of conditional gradient algorithms, notably avoidance of projections onto P and maintenance of iterates as sparse convex combinations of a limited number of extreme points of P. The algorithm is lazy, making use of inexpensive inexact solutions of the linear programming subproblem that characterizes the conditional gradient approach. It decreases measures of optimality (primal and dual gaps) rapidly, both in the number of iterations and in wall-clock time, outperforming even the lazy conditional gradient algorithms of Braun et al. 2017. We also present a streamlined version of the algorithm that applies when P is the probability simplex.'
volume: 97
URL: https://proceedings.mlr.press/v97/braun19a.html
PDF: http://proceedings.mlr.press/v97/braun19a/braun19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-braun19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Gábor
family: Braun
- given: Sebastian
family: Pokutta
- given: Dan
family: Tu
- given: Stephen
family: Wright
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 735-743
id: braun19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 735
lastpage: 743
published: 2019-05-24 00:00:00 +0000
- title: 'Coresets for Ordered Weighted Clustering'
abstract: 'We design coresets for Ordered k-Median, a generalization of classical clustering problems such as k-Median and k-Center. Its objective function is defined via the Ordered Weighted Averaging (OWA) paradigm of Yager (1988), where data points are weighted according to a predefined weight vector, but in order of their contribution to the objective (distance from the centers). A powerful data-reduction technique, called a coreset, is to summarize a point set $X$ in $\mathbb{R}^d$ into a small (weighted) point set $X’$, such that for every set of $k$ potential centers, the objective value of the coreset $X’$ approximates that of $X$ within factor $1\pm \epsilon$. When there are multiple objectives (weights), the above standard coreset might have limited usefulness, whereas in a *simultaneous* coreset, the above approximation holds for all weights (in addition to all centers). Our main result is a construction of a simultaneous coreset of size $O_{\epsilon, d}(k^2 \log^2 |X|)$ for Ordered k-Median. We validate our algorithm on a real geographical data set, and we find our coreset leads to a massive speedup of clustering computations, while maintaining high accuracy for a range of weights.'
volume: 97
URL: https://proceedings.mlr.press/v97/braverman19a.html
PDF: http://proceedings.mlr.press/v97/braverman19a/braverman19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-braverman19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Vladimir
family: Braverman
- given: Shaofeng H.-C.
family: Jiang
- given: Robert
family: Krauthgamer
- given: Xuan
family: Wu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 744-753
id: braverman19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 744
lastpage: 753
published: 2019-05-24 00:00:00 +0000
- title: 'Target Tracking for Contextual Bandits: Application to Demand Side Management'
abstract: 'We propose a contextual-bandit approach for demand side management by offering price incentives. More precisely, a target mean consumption is set at each round and the mean consumption is modeled as a complex function of the distribution of prices sent and of some contextual variables such as the temperature, weather, and so on. The performance of our strategies is measured in quadratic losses through a regret criterion. We offer $T^{2/3}$ upper bounds on this regret (up to poly-logarithmic terms)—and even faster rates under stronger assumptions—for strategies inspired by standard strategies for contextual bandits (like LinUCB, see Li et al., 2010). Simulations on a real data set gathered by UK Power Networks, in which price incentives were offered, show that our strategies are effective and may indeed manage demand response by suitably picking the price levels.'
volume: 97
URL: https://proceedings.mlr.press/v97/bregere19a.html
PDF: http://proceedings.mlr.press/v97/bregere19a/bregere19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-bregere19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Margaux
family: Brégère
- given: Pierre
family: Gaillard
- given: Yannig
family: Goude
- given: Gilles
family: Stoltz
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 754-763
id: bregere19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 754
lastpage: 763
published: 2019-05-24 00:00:00 +0000
- title: 'Active Manifolds: A non-linear analogue to Active Subspaces'
abstract: 'We present an approach to analyze $C^1(\mathbb{R}^m)$ functions that addresses limitations present in the Active Subspaces (AS) method of Constantine et al. (2014; 2015). Under appropriate hypotheses, our Active Manifolds (AM) method identifies a 1-D curve in the domain (the active manifold) on which nearly all values of the unknown function are attained, which can be exploited for approximation or analysis, especially when $m$ is large (high-dimensional input space). We provide theorems justifying our AM technique and an algorithm permitting functional approximation and sensitivity analysis. Using accessible, low-dimensional functions as initial examples, we show AM reduces approximation error by an order of magnitude compared to AS, at the expense of more computation. Following this, we revisit the sensitivity analysis by Glaws et al. (2017), who apply AS to analyze a magnetohydrodynamic power generator model, and compare the performance of AM on the same data. Our analysis provides detailed information not captured by AS, exhibiting the influence of each parameter individually along an active manifold. Overall, AM represents a novel technique for analyzing functional models with benefits including: reducing $m$-dimensional analysis to a 1-D analogue, permitting more accurate regression than AS (at more computational expense), enabling more informative sensitivity analysis, and granting accessible visualizations (2-D plots) of parameter sensitivity along the AM.'
volume: 97
URL: https://proceedings.mlr.press/v97/bridges19a.html
PDF: http://proceedings.mlr.press/v97/bridges19a/bridges19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-bridges19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Robert
family: Bridges
- given: Anthony
family: Gruber
- given: Christopher
family: Felder
- given: Miki
family: Verma
- given: Chelsey
family: Hoff
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 764-772
id: bridges19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 764
lastpage: 772
published: 2019-05-24 00:00:00 +0000
- title: 'Conditioning by adaptive sampling for robust design'
abstract: 'We present a method for design problems wherein the goal is to maximize or specify the value of one or more properties of interest (e.g. maximizing the fluorescence of a protein). We assume access to black box, stochastic “oracle" predictive functions, each of which maps from design space to a distribution over properties of interest. Because many state-of-the-art predictive models are known to suffer from pathologies, especially for data far from the training distribution, the problem becomes different from directly optimizing the oracles. Herein, we propose a method to solve this problem that uses model-based adaptive sampling to estimate a distribution over the design space, conditioned on the desired properties.'
volume: 97
URL: https://proceedings.mlr.press/v97/brookes19a.html
PDF: http://proceedings.mlr.press/v97/brookes19a/brookes19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-brookes19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: David
family: Brookes
- given: Hahnbeom
family: Park
- given: Jennifer
family: Listgarten
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 773-782
id: brookes19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 773
lastpage: 782
published: 2019-05-24 00:00:00 +0000
- title: 'Extrapolating Beyond Suboptimal Demonstrations via Inverse Reinforcement Learning from Observations'
abstract: 'A critical flaw of existing inverse reinforcement learning (IRL) methods is their inability to significantly outperform the demonstrator. This is because IRL typically seeks a reward function that makes the demonstrator appear near-optimal, rather than inferring the underlying intentions of the demonstrator that may have been poorly executed in practice. In this paper, we introduce a novel reward-learning-from-observation algorithm, Trajectory-ranked Reward EXtrapolation (T-REX), that extrapolates beyond a set of (approximately) ranked demonstrations in order to infer high-quality reward functions from a set of potentially poor demonstrations. When combined with deep reinforcement learning, T-REX outperforms state-of-the-art imitation learning and IRL methods on multiple Atari and MuJoCo benchmark tasks and achieves performance that is often more than twice the performance of the best demonstration. We also demonstrate that T-REX is robust to ranking noise and can accurately extrapolate intention by simply watching a learner noisily improve at a task over time.'
volume: 97
URL: https://proceedings.mlr.press/v97/brown19a.html
PDF: http://proceedings.mlr.press/v97/brown19a/brown19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-brown19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Daniel
family: Brown
- given: Wonjoon
family: Goo
- given: Prabhat
family: Nagarajan
- given: Scott
family: Niekum
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 783-792
id: brown19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 783
lastpage: 792
published: 2019-05-24 00:00:00 +0000
- title: 'Deep Counterfactual Regret Minimization'
abstract: 'Counterfactual Regret Minimization (CFR) is the leading algorithm for solving large imperfect-information games. It converges to an equilibrium by iteratively traversing the game tree. In order to deal with extremely large games, abstraction is typically applied before running CFR. The abstracted game is solved with tabular CFR, and its solution is mapped back to the full game. This process can be problematic because aspects of abstraction are often manual and domain specific, abstraction algorithms may miss important strategic nuances of the game, and there is a chicken-and-egg problem because determining a good abstraction requires knowledge of the equilibrium of the game. This paper introduces *Deep Counterfactual Regret Minimization*, a form of CFR that obviates the need for abstraction by instead using deep neural networks to approximate the behavior of CFR in the full game. We show that Deep CFR is principled and achieves strong performance in large poker games. This is the first non-tabular variant of CFR to be successful in large games.'
volume: 97
URL: https://proceedings.mlr.press/v97/brown19b.html
PDF: http://proceedings.mlr.press/v97/brown19b/brown19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-brown19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Noam
family: Brown
- given: Adam
family: Lerer
- given: Sam
family: Gross
- given: Tuomas
family: Sandholm
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 793-802
id: brown19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 793
lastpage: 802
published: 2019-05-24 00:00:00 +0000
- title: 'Understanding the Origins of Bias in Word Embeddings'
abstract: 'Popular word embedding algorithms exhibit stereotypical biases, such as gender bias. The widespread use of these algorithms in machine learning systems can amplify stereotypes in important contexts. Although some methods have been developed to mitigate this problem, how word embedding biases arise during training is poorly understood. In this work we develop a technique to address this question. Given a word embedding, our method reveals how perturbing the training corpus would affect the resulting embedding bias. By tracing the origins of word embedding bias back to the original training documents, one can identify subsets of documents whose removal would most reduce bias. We demonstrate our methodology on Wikipedia and New York Times corpora, and find it to be very accurate.'
volume: 97
URL: https://proceedings.mlr.press/v97/brunet19a.html
PDF: http://proceedings.mlr.press/v97/brunet19a/brunet19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-brunet19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Marc-Etienne
family: Brunet
- given: Colleen
family: Alkalay-Houlihan
- given: Ashton
family: Anderson
- given: Richard
family: Zemel
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 803-811
id: brunet19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 803
lastpage: 811
published: 2019-05-24 00:00:00 +0000
- title: 'Low Latency Privacy Preserving Inference'
abstract: 'When applying machine learning to sensitive data, one has to find a balance between accuracy, information security, and computational-complexity. Recent studies combined Homomorphic Encryption with neural networks to make inferences while protecting against information leakage. However, these methods are limited by the width and depth of neural networks that can be used (and hence the accuracy) and exhibit high latency even for relatively simple networks. In this study we provide two solutions that address these limitations. In the first solution, we present more than 10\times improvement in latency and enable inference on wider networks compared to prior attempts with the same level of security. The improved performance is achieved by novel methods to represent the data during the computation. In the second solution, we apply the method of transfer learning to provide private inference services using deep networks with latency of \sim0.16 seconds. We demonstrate the efficacy of our methods on several computer vision tasks.'
volume: 97
URL: https://proceedings.mlr.press/v97/brutzkus19a.html
PDF: http://proceedings.mlr.press/v97/brutzkus19a/brutzkus19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-brutzkus19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alon
family: Brutzkus
- given: Ran
family: Gilad-Bachrach
- given: Oren
family: Elisha
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 812-821
id: brutzkus19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 812
lastpage: 821
published: 2019-05-24 00:00:00 +0000
- title: 'Why do Larger Models Generalize Better? A Theoretical Perspective via the XOR Problem'
abstract: 'Empirical evidence suggests that neural networks with ReLU activations generalize better with over-parameterization. However, there is currently no theoretical analysis that explains this observation. In this work, we provide theoretical and empirical evidence that, in certain cases, overparameterized convolutional networks generalize better than small networks because of an interplay between weight clustering and feature exploration at initialization. We demonstrate this theoretically for a 3-layer convolutional neural network with max-pooling, in a novel setting which extends the XOR problem. We show that this interplay implies that with overparamterization, gradient descent converges to global minima with better generalization performance compared to global minima of small networks. Empirically, we demonstrate these phenomena for a 3-layer convolutional neural network in the MNIST task.'
volume: 97
URL: https://proceedings.mlr.press/v97/brutzkus19b.html
PDF: http://proceedings.mlr.press/v97/brutzkus19b/brutzkus19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-brutzkus19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alon
family: Brutzkus
- given: Amir
family: Globerson
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 822-830
id: brutzkus19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 822
lastpage: 830
published: 2019-05-24 00:00:00 +0000
- title: 'Adversarial examples from computational constraints'
abstract: 'Why are classifiers in high dimension vulnerable to “adversarial” perturbations? We show that it is likely not due to information theoretic limitations, but rather it could be due to computational constraints. First we prove that, for a broad set of classification tasks, the mere existence of a robust classifier implies that it can be found by a possibly exponential-time algorithm with relatively few training examples. Then we give two particular classification tasks where learning a robust classifier is computationally intractable. More precisely we construct two binary classifications task in high dimensional space which are (i) information theoretically easy to learn robustly for large perturbations, (ii) efficiently learnable (non-robustly) by a simple linear separator, (iii) yet are not efficiently robustly learnable, even for small perturbations. Specifically, for the first task hardness holds for any efficient algorithm in the statistical query (SQ) model, while for the second task we rule out any efficient algorithm under a cryptographic assumption. These examples give an exponential separation between classical learning and robust learning in the statistical query model or under a cryptographic assumption. It suggests that adversarial examples may be an unavoidable byproduct of computational limitations of learning algorithms.'
volume: 97
URL: https://proceedings.mlr.press/v97/bubeck19a.html
PDF: http://proceedings.mlr.press/v97/bubeck19a/bubeck19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-bubeck19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sebastien
family: Bubeck
- given: Yin Tat
family: Lee
- given: Eric
family: Price
- given: Ilya
family: Razenshteyn
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 831-840
id: bubeck19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 831
lastpage: 840
published: 2019-05-24 00:00:00 +0000
- title: 'Self-similar Epochs: Value in arrangement'
abstract: 'Optimization of machine learning models is commonly performed through stochastic gradient updates on randomly ordered training examples. This practice means that each fraction of an epoch comprises an independent random sample of the training data that may not preserve informative structure present in the full data. We hypothesize that the training can be more effective with *self-similar* arrangements that potentially allow each epoch to provide benefits of multiple ones. We study this for “matrix factorization” – the common task of learning metric embeddings of entities such as queries, videos, or words from example pairwise associations. We construct arrangements that preserve the weighted Jaccard similarities of rows and columns and experimentally observe training acceleration of 3%-37% on synthetic and recommendation datasets. Principled arrangements of training examples emerge as a novel and potentially powerful enhancement to SGD that merits further exploration.'
volume: 97
URL: https://proceedings.mlr.press/v97/buchnik19a.html
PDF: http://proceedings.mlr.press/v97/buchnik19a/buchnik19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-buchnik19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Eliav
family: Buchnik
- given: Edith
family: Cohen
- given: Avinatan
family: Hasidim
- given: Yossi
family: Matias
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 841-850
id: buchnik19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 841
lastpage: 850
published: 2019-05-24 00:00:00 +0000
- title: 'Learning Generative Models across Incomparable Spaces'
abstract: 'Generative Adversarial Networks have shown remarkable success in learning a distribution that faithfully recovers a reference distribution in its entirety. However, in some cases, we may want to only learn some aspects (e.g., cluster or manifold structure), while modifying others (e.g., style, orientation or dimension). In this work, we propose an approach to learn generative models across such incomparable spaces, and demonstrate how to steer the learned distribution towards target properties. A key component of our model is the Gromov-Wasserstein distance, a notion of discrepancy that compares distributions relationally rather than absolutely. While this framework subsumes current generative models in identically reproducing distributions, its inherent flexibility allows application to tasks in manifold learning, relational learning and cross-domain learning.'
volume: 97
URL: https://proceedings.mlr.press/v97/bunne19a.html
PDF: http://proceedings.mlr.press/v97/bunne19a/bunne19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-bunne19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Charlotte
family: Bunne
- given: David
family: Alvarez-Melis
- given: Andreas
family: Krause
- given: Stefanie
family: Jegelka
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 851-861
id: bunne19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 851
lastpage: 861
published: 2019-05-24 00:00:00 +0000
- title: 'Rates of Convergence for Sparse Variational Gaussian Process Regression'
abstract: 'Excellent variational approximations to Gaussian process posteriors have been developed which avoid the $\mathcal{O}\left(N^3\right)$ scaling with dataset size $N$. They reduce the computational cost to $\mathcal{O}\left(NM^2\right)$, with $M\ll N$ the number of *inducing variables*, which summarise the process. While the computational cost seems to be linear in $N$, the true complexity of the algorithm depends on how $M$ must increase to ensure a certain quality of approximation. We show that with high probability the KL divergence can be made arbitrarily small by growing $M$ more slowly than $N$. A particular case is that for regression with normally distributed inputs in D-dimensions with the Squared Exponential kernel, $M=\mathcal{O}(\log^D N)$ suffices. Our results show that as datasets grow, Gaussian process posteriors can be approximated cheaply, and provide a concrete rule for how to increase $M$ in continual learning scenarios.'
volume: 97
URL: https://proceedings.mlr.press/v97/burt19a.html
PDF: http://proceedings.mlr.press/v97/burt19a/burt19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-burt19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: David
family: Burt
- given: Carl Edward
family: Rasmussen
- given: Mark
family: Van Der Wilk
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 862-871
id: burt19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 862
lastpage: 871
published: 2019-05-24 00:00:00 +0000
- title: 'What is the Effect of Importance Weighting in Deep Learning?'
abstract: 'Importance-weighted risk minimization is a key ingredient in many machine learning algorithms for causal inference, domain adaptation, class imbalance, and off-policy reinforcement learning. While the effect of importance weighting is well-characterized for low-capacity misspecified models, little is known about how it impacts over-parameterized, deep neural networks. This work is inspired by recent theoretical results showing that on (linearly) separable data, deep linear networks optimized by SGD learn weight-agnostic solutions, prompting us to ask, for realistic deep networks, for which many practical datasets are separable, what is the effect of importance weighting? We present the surprising finding that while importance weighting impacts models early in training, its effect diminishes over successive epochs. Moreover, while L2 regularization and batch normalization (but not dropout), restore some of the impact of importance weighting, they express the effect via (seemingly) the wrong abstraction: why should practitioners tweak the L2 regularization, and by how much, to produce the correct weighting effect? Our experiments confirm these findings across a range of architectures and datasets.'
volume: 97
URL: https://proceedings.mlr.press/v97/byrd19a.html
PDF: http://proceedings.mlr.press/v97/byrd19a/byrd19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-byrd19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jonathon
family: Byrd
- given: Zachary
family: Lipton
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 872-881
id: byrd19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 872
lastpage: 881
published: 2019-05-24 00:00:00 +0000
- title: 'A Quantitative Analysis of the Effect of Batch Normalization on Gradient Descent'
abstract: 'Despite its empirical success and recent theoretical progress, there generally lacks a quantitative analysis of the effect of batch normalization (BN) on the convergence and stability of gradient descent. In this paper, we provide such an analysis on the simple problem of ordinary least squares (OLS), where the precise dynamical properties of gradient descent (GD) is completely known, thus allowing us to isolate and compare the additional effects of BN. More precisely, we show that unlike GD, gradient descent with BN (BNGD) converges for arbitrary learning rates for the weights, and the convergence remains linear under mild conditions. Moreover, we quantify two different sources of acceleration of BNGD over GD – one due to over-parameterization which improves the effective condition number and another due having a large range of learning rates giving rise to fast descent. These phenomena set BNGD apart from GD and could account for much of its robustness properties. These findings are confirmed quantitatively by numerical experiments, which further show that many of the uncovered properties of BNGD in OLS are also observed qualitatively in more complex supervised learning problems.'
volume: 97
URL: https://proceedings.mlr.press/v97/cai19a.html
PDF: http://proceedings.mlr.press/v97/cai19a/cai19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cai19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yongqiang
family: Cai
- given: Qianxiao
family: Li
- given: Zuowei
family: Shen
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 882-890
id: cai19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 882
lastpage: 890
published: 2019-05-24 00:00:00 +0000
- title: 'Accelerated Linear Convergence of Stochastic Momentum Methods in Wasserstein Distances'
abstract: 'Momentum methods such as Polyak’s heavy ball (HB) method, Nesterov’s accelerated gradient (AG) as well as accelerated projected gradient (APG) method have been commonly used in machine learning practice, but their performance is quite sensitive to noise in the gradients. We study these methods under a first-order stochastic oracle model where noisy estimates of the gradients are available. For strongly convex problems, we show that the distribution of the iterates of AG converges with the accelerated $O(\sqrt{\kappa}\log(1/\varepsilon))$ linear rate to a ball of radius $\varepsilon$ centered at a unique invariant distribution in the 1-Wasserstein metric where $\kappa$ is the condition number as long as the noise variance is smaller than an explicit upper bound we can provide. Our analysis also certifies linear convergence rates as a function of the stepsize, momentum parameter and the noise variance; recovering the accelerated rates in the noiseless case and quantifying the level of noise that can be tolerated to achieve a given performance. To the best of our knowledge, these are the first linear convergence results for stochastic momentum methods under the stochastic oracle model. We also develop finer results for the special case of quadratic objectives, extend our results to the APG method and weakly convex functions showing accelerated rates when the noise magnitude is sufficiently small.'
volume: 97
URL: https://proceedings.mlr.press/v97/can19a.html
PDF: http://proceedings.mlr.press/v97/can19a/can19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-can19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Bugra
family: Can
- given: Mert
family: Gurbuzbalaban
- given: Lingjiong
family: Zhu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 891-901
id: can19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 891
lastpage: 901
published: 2019-05-24 00:00:00 +0000
- title: 'Active Embedding Search via Noisy Paired Comparisons'
abstract: 'Suppose that we wish to estimate a user’s preference vector $w$ from paired comparisons of the form “does user $w$ prefer item $p$ or item $q$?,” where both the user and items are embedded in a low-dimensional Euclidean space with distances that reflect user and item similarities. Such observations arise in numerous settings, including psychometrics and psychology experiments, search tasks, advertising, and recommender systems. In such tasks, queries can be extremely costly and subject to varying levels of response noise; thus, we aim to actively choose pairs that are most informative given the results of previous comparisons. We provide new theoretical insights into the benefits and challenges of greedy information maximization in this setting, and develop two novel strategies that maximize lower bounds on information gain and are simpler to analyze and compute respectively. We use simulated responses from a real-world dataset to validate our strategies through their similar performance to greedy information maximization, and their superior preference estimation over state-of-the-art selection methods as well as random queries.'
volume: 97
URL: https://proceedings.mlr.press/v97/canal19a.html
PDF: http://proceedings.mlr.press/v97/canal19a/canal19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-canal19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Gregory
family: Canal
- given: Andy
family: Massimino
- given: Mark
family: Davenport
- given: Christopher
family: Rozell
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 902-911
id: canal19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 902
lastpage: 911
published: 2019-05-24 00:00:00 +0000
- title: 'Dynamic Learning with Frequent New Product Launches: A Sequential Multinomial Logit Bandit Problem'
abstract: 'Motivated by the phenomenon that companies introduce new products to keep abreast with customers’ rapidly changing tastes, we consider a novel online learning setting where a profit-maximizing seller needs to learn customers’ preferences through offering recommendations, which may contain existing products and new products that are launched in the middle of a selling period. We propose a sequential multinomial logit (SMNL) model to characterize customers’ behavior when product recommendations are presented in tiers. For the offline version with known customers’ preferences, we propose a polynomial-time algorithm and characterize the properties of the optimal tiered product recommendation. For the online problem, we propose a learning algorithm and quantify its regret bound. Moreover, we extend the setting to incorporate a constraint which ensures every new product is learned to a given accuracy. Our results demonstrate the tier structure can be used to mitigate the risks associated with learning new products.'
volume: 97
URL: https://proceedings.mlr.press/v97/cao19a.html
PDF: http://proceedings.mlr.press/v97/cao19a/cao19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cao19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Junyu
family: Cao
- given: Wei
family: Sun
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 912-920
id: cao19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 912
lastpage: 920
published: 2019-05-24 00:00:00 +0000
- title: 'Competing Against Nash Equilibria in Adversarially Changing Zero-Sum Games'
abstract: 'We study the problem of repeated play in a zero-sum game in which the payoff matrix may change, in a possibly adversarial fashion, on each round; we call these Online Matrix Games. Finding the Nash Equilibrium (NE) of a two player zero-sum game is core to many problems in statistics, optimization, and economics, and for a fixed game matrix this can be easily reduced to solving a linear program. But when the payoff matrix evolves over time our goal is to find a sequential algorithm that can compete with, in a certain sense, the NE of the long-term-averaged payoff matrix. We design an algorithm with small NE regret–that is, we ensure that the long-term payoff of both players is close to minimax optimum in hindsight. Our algorithm achieves near-optimal dependence with respect to the number of rounds and depends poly-logarithmically on the number of available actions of the players. Additionally, we show that the naive reduction, where each player simply minimizes its own regret, fails to achieve the stated objective regardless of which algorithm is used. Lastly, we consider the so-called bandit setting, where the feedback is significantly limited, and we provide an algorithm with small NE regret using one-point estimates of each payoff matrix.'
volume: 97
URL: https://proceedings.mlr.press/v97/cardoso19a.html
PDF: http://proceedings.mlr.press/v97/cardoso19a/cardoso19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cardoso19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Adrian Rivera
family: Cardoso
- given: Jacob
family: Abernethy
- given: He
family: Wang
- given: Huan
family: Xu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 921-930
id: cardoso19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 921
lastpage: 930
published: 2019-05-24 00:00:00 +0000
- title: 'Automated Model Selection with Bayesian Quadrature'
abstract: 'We present a novel technique for tailoring Bayesian quadrature (BQ) to model selection. The state-of-the-art for comparing the evidence of multiple models relies on Monte Carlo methods, which converge slowly and are unreliable for computationally expensive models. Although previous research has shown that BQ offers sample efficiency superior to Monte Carlo in computing the evidence of an individual model, applying BQ directly to model comparison may waste computation producing an overly-accurate estimate for the evidence of a clearly poor model. We propose an automated and efficient algorithm for computing the most-relevant quantity for model selection: the posterior model probability. Our technique maximizes the mutual information between this quantity and observations of the models’ likelihoods, yielding efficient sample acquisition across disparate model spaces when likelihood observations are limited. Our method produces more-accurate posterior estimates using fewer likelihood evaluations than standard Bayesian quadrature and Monte Carlo estimators, as we demonstrate on synthetic and real-world examples.'
volume: 97
URL: https://proceedings.mlr.press/v97/chai19a.html
PDF: http://proceedings.mlr.press/v97/chai19a/chai19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chai19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Henry
family: Chai
- given: Jean-Francois
family: Ton
- given: Michael A.
family: Osborne
- given: Roman
family: Garnett
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 931-940
id: chai19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 931
lastpage: 940
published: 2019-05-24 00:00:00 +0000
- title: 'Learning Action Representations for Reinforcement Learning'
abstract: 'Most model-free reinforcement learning methods leverage state representations (embeddings) for generalization, but either ignore structure in the space of actions or assume the structure is provided a priori. We show how a policy can be decomposed into a component that acts in a low-dimensional space of action representations and a component that transforms these representations into actual actions. These representations improve generalization over large, finite action sets by allowing the agent to infer the outcomes of actions similar to actions already taken. We provide an algorithm to both learn and use action representations and provide conditions for its convergence. The efficacy of the proposed method is demonstrated on large-scale real-world problems.'
volume: 97
URL: https://proceedings.mlr.press/v97/chandak19a.html
PDF: http://proceedings.mlr.press/v97/chandak19a/chandak19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chandak19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yash
family: Chandak
- given: Georgios
family: Theocharous
- given: James
family: Kostas
- given: Scott
family: Jordan
- given: Philip
family: Thomas
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 941-950
id: chandak19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 941
lastpage: 950
published: 2019-05-24 00:00:00 +0000
- title: 'Dynamic Measurement Scheduling for Event Forecasting using Deep RL'
abstract: 'Imagine a patient in critical condition. What and when should be measured to forecast detrimental events, especially under the budget constraints? We answer this question by deep reinforcement learning (RL) that jointly minimizes the measurement cost and maximizes predictive gain, by scheduling strategically-timed measurements. We learn our policy to be dynamically dependent on the patient’s health history. To scale our framework to exponentially large action space, we distribute our reward in a sequential setting that makes the learning easier. In our simulation, our policy outperforms heuristic-based scheduling with higher predictive gain and lower cost. In a real-world ICU mortality prediction task (MIMIC3), our policies reduce the total number of measurements by 31% or improve predictive gain by a factor of 3 as compared to physicians, under the off-policy policy evaluation.'
volume: 97
URL: https://proceedings.mlr.press/v97/chang19a.html
PDF: http://proceedings.mlr.press/v97/chang19a/chang19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chang19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Chun-Hao
family: Chang
- given: Mingjie
family: Mai
- given: Anna
family: Goldenberg
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 951-960
id: chang19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 951
lastpage: 960
published: 2019-05-24 00:00:00 +0000
- title: 'On Symmetric Losses for Learning from Corrupted Labels'
abstract: 'This paper aims to provide a better understanding of a symmetric loss. First, we emphasize that using a symmetric loss is advantageous in the balanced error rate (BER) minimization and area under the receiver operating characteristic curve (AUC) maximization from corrupted labels. Second, we prove general theoretical properties of symmetric losses, including a classification-calibration condition, excess risk bound, conditional risk minimizer, and AUC-consistency condition. Third, since all nonnegative symmetric losses are non-convex, we propose a convex barrier hinge loss that benefits significantly from the symmetric condition, although it is not symmetric everywhere. Finally, we conduct experiments to validate the relevance of the symmetric condition.'
volume: 97
URL: https://proceedings.mlr.press/v97/charoenphakdee19a.html
PDF: http://proceedings.mlr.press/v97/charoenphakdee19a/charoenphakdee19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-charoenphakdee19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Nontawat
family: Charoenphakdee
- given: Jongyeong
family: Lee
- given: Masashi
family: Sugiyama
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 961-970
id: charoenphakdee19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 961
lastpage: 970
published: 2019-05-24 00:00:00 +0000
- title: 'Online learning with kernel losses'
abstract: 'We present a generalization of the adversarial linear bandits framework, where the underlying losses are kernel functions (with an associated reproducing kernel Hilbert space) rather than linear functions. We study a version of the exponential weights algorithm and bound its regret in this setting. Under conditions on the eigen-decay of the kernel we provide a sharp characterization of the regret for this algorithm. When we have polynomial eigen-decay ($\mu_j \le \mathcal{O}(j^{-\beta})$), we find that the regret is bounded by $\mathcal{R}_n \le \mathcal{O}(n^{\beta/2(\beta-1)})$. While under the assumption of exponential eigen-decay ($\mu_j \le \mathcal{O}(e^{-\beta j })$) we get an even tighter bound on the regret $\mathcal{R}_n \le \tilde{\mathcal{O}}(n^{1/2})$. When the eigen-decay is polynomial we also show a *non-matching* minimax lower bound on the regret of $\mathcal{R}_n \ge \Omega(n^{(\beta+1)/2\beta})$ and a lower bound of $\mathcal{R}_n \ge \Omega(n^{1/2})$ when the decay in the eigen-values is exponentially fast. We also study the full information setting when the underlying losses are kernel functions and present an adapted exponential weights algorithm and a conditional gradient descent algorithm.'
volume: 97
URL: https://proceedings.mlr.press/v97/chatterji19a.html
PDF: http://proceedings.mlr.press/v97/chatterji19a/chatterji19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chatterji19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Niladri
family: Chatterji
- given: Aldo
family: Pacchiano
- given: Peter
family: Bartlett
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 971-980
id: chatterji19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 971
lastpage: 980
published: 2019-05-24 00:00:00 +0000
- title: 'Neural Network Attributions: A Causal Perspective'
abstract: 'We propose a new attribution method for neural networks developed using ﬁrst principles of causality (to the best of our knowledge, the ﬁrst such). The neural network architecture is viewed as a Structural Causal Model, and a methodology to compute the causal effect of each feature on the output is presented. With reasonable assumptions on the causal structure of the input data, we propose algorithms to efﬁciently compute the causal effects, as well as scale the approach to data with large dimensionality. We also show how this method can be used for recurrent neural networks. We report experimental results on both simulated and real datasets showcasing the promise and usefulness of the proposed algorithm.'
volume: 97
URL: https://proceedings.mlr.press/v97/chattopadhyay19a.html
PDF: http://proceedings.mlr.press/v97/chattopadhyay19a/chattopadhyay19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chattopadhyay19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Aditya
family: Chattopadhyay
- given: Piyushi
family: Manupriya
- given: Anirban
family: Sarkar
- given: Vineeth N
family: Balasubramanian
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 981-990
id: chattopadhyay19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 981
lastpage: 990
published: 2019-05-24 00:00:00 +0000
- title: 'PAC Identification of Many Good Arms in Stochastic Multi-Armed Bandits'
abstract: 'We consider the problem of identifying any k out of the best m arms in an n-armed stochastic multi-armed bandit; framed in the PAC setting, this particular problem generalises both the problem of “best subset selection” (Kalyanakrishnan & Stone, 2010) and that of selecting “one out of the best m” arms (Roy Chaudhuri & Kalyanakrishnan, 2017). We present a lower bound on the worst-case sample complexity for general k, and a fully sequential PAC algorithm, LUCB-k-m, which is more sample-efficient on easy instances. Also, extending our analysis to infinite-armed bandits, we present a PAC algorithm that is independent of n, which identifies an arm from the best $\rho$ fraction of arms using at most an additive poly-log number of samples than compared to the lower bound, thereby improving over Roy Chaudhuri & Kalyanakrishnan (2017) and Aziz et al. (2018). The problem of identifying k > 1 distinct arms from the best $\rho$ fraction is not always well-defined; for a special class of this problem, we present lower and upper bounds. Finally, through a reduction, we establish a relation between upper bounds for the “one out of the best $\rho$” problem for infinite instances and the “one out of the best m” problem for finite instances. We conjecture that it is more efficient to solve “small” finite instances using the latter formulation, rather than going through the former.'
volume: 97
URL: https://proceedings.mlr.press/v97/chaudhuri19a.html
PDF: http://proceedings.mlr.press/v97/chaudhuri19a/chaudhuri19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chaudhuri19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Arghya Roy
family: Chaudhuri
- given: Shivaram
family: Kalyanakrishnan
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 991-1000
id: chaudhuri19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 991
lastpage: 1000
published: 2019-05-24 00:00:00 +0000
- title: 'Nearest Neighbor and Kernel Survival Analysis: Nonasymptotic Error Bounds and Strong Consistency Rates'
abstract: 'We establish the first nonasymptotic error bounds for Kaplan-Meier-based nearest neighbor and kernel survival probability estimators where feature vectors reside in metric spaces. Our bounds imply rates of strong consistency for these nonparametric estimators and, up to a log factor, match an existing lower bound for conditional CDF estimation. Our proof strategy also yields nonasymptotic guarantees for nearest neighbor and kernel variants of the Nelson-Aalen cumulative hazards estimator. We experimentally compare these methods on four datasets. We find that for the kernel survival estimator, a good choice of kernel is one learned using random survival forests.'
volume: 97
URL: https://proceedings.mlr.press/v97/chen19a.html
PDF: http://proceedings.mlr.press/v97/chen19a/chen19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chen19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: George
family: Chen
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1001-1010
id: chen19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1001
lastpage: 1010
published: 2019-05-24 00:00:00 +0000
- title: 'Stein Point Markov Chain Monte Carlo'
abstract: 'An important task in machine learning and statistics is the approximation of a probability measure by an empirical measure supported on a discrete point set. Stein Points are a class of algorithms for this task, which proceed by sequentially minimising a Stein discrepancy between the empirical measure and the target and, hence, require the solution of a non-convex optimisation problem to obtain each new point. This paper removes the need to solve this optimisation problem by, instead, selecting each new point based on a Markov chain sample path. This significantly reduces the computational cost of Stein Points and leads to a suite of algorithms that are straightforward to implement. The new algorithms are illustrated on a set of challenging Bayesian inference problems, and rigorous theoretical guarantees of consistency are established.'
volume: 97
URL: https://proceedings.mlr.press/v97/chen19b.html
PDF: http://proceedings.mlr.press/v97/chen19b/chen19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chen19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Wilson Ye
family: Chen
- given: Alessandro
family: Barp
- given: Francois-Xavier
family: Briol
- given: Jackson
family: Gorham
- given: Mark
family: Girolami
- given: Lester
family: Mackey
- given: Chris
family: Oates
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1011-1021
id: chen19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1011
lastpage: 1021
published: 2019-05-24 00:00:00 +0000
- title: 'Particle Flow Bayes’ Rule'
abstract: 'We present a particle flow realization of Bayes’ rule, where an ODE-based neural operator is used to transport particles from a prior to its posterior after a new observation. We prove that such an ODE operator exists. Its neural parameterization can be trained in a meta-learning framework, allowing this operator to reason about the effect of an individual observation on the posterior, and thus generalize across different priors, observations and to sequential Bayesian inference. We demonstrated the generalization ability of our particle flow Bayes operator in several canonical and high dimensional examples.'
volume: 97
URL: https://proceedings.mlr.press/v97/chen19c.html
PDF: http://proceedings.mlr.press/v97/chen19c/chen19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chen19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Xinshi
family: Chen
- given: Hanjun
family: Dai
- given: Le
family: Song
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1022-1031
id: chen19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1022
lastpage: 1031
published: 2019-05-24 00:00:00 +0000
- title: 'Proportionally Fair Clustering'
abstract: 'We extend the fair machine learning literature by considering the problem of proportional centroid clustering in a metric context. For clustering n points with k centers, we define fairness as proportionality to mean that any n/k points are entitled to form their own cluster if there is another center that is closer in distance for all n/k points. We seek clustering solutions to which there are no such justified complaints from any subsets of agents, without assuming any a priori notion of protected subsets. We present and analyze algorithms to efficiently compute, optimize, and audit proportional solutions. We conclude with an empirical examination of the tradeoff between proportional solutions and the k-means objective.'
volume: 97
URL: https://proceedings.mlr.press/v97/chen19d.html
PDF: http://proceedings.mlr.press/v97/chen19d/chen19d.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chen19d.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Xingyu
family: Chen
- given: Brandon
family: Fain
- given: Liang
family: Lyu
- given: Kamesh
family: Munagala
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1032-1041
id: chen19d
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1032
lastpage: 1041
published: 2019-05-24 00:00:00 +0000
- title: 'Information-Theoretic Considerations in Batch Reinforcement Learning'
abstract: 'Value-function approximation methods that operate in batch mode have foundational importance to reinforcement learning (RL). Finite sample guarantees for these methods often crucially rely on two types of assumptions: (1) mild distribution shift, and (2) representation conditions that are stronger than realizability. However, the necessity (“why do we need them?”) and the naturalness (“when do they hold?”) of such assumptions have largely eluded the literature. In this paper, we revisit these assumptions and provide theoretical results towards answering the above questions, and make steps towards a deeper understanding of value-function approximation.'
volume: 97
URL: https://proceedings.mlr.press/v97/chen19e.html
PDF: http://proceedings.mlr.press/v97/chen19e/chen19e.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chen19e.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jinglin
family: Chen
- given: Nan
family: Jiang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1042-1051
id: chen19e
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1042
lastpage: 1051
published: 2019-05-24 00:00:00 +0000
- title: 'Generative Adversarial User Model for Reinforcement Learning Based Recommendation System'
abstract: 'There are great interests as well as many challenges in applying reinforcement learning (RL) to recommendation systems. In this setting, an online user is the environment; neither the reward function nor the environment dynamics are clearly defined, making the application of RL challenging. In this paper, we propose a novel model-based reinforcement learning framework for recommendation systems, where we develop a generative adversarial network to imitate user behavior dynamics and learn her reward function. Using this user model as the simulation environment, we develop a novel Cascading DQN algorithm to obtain a combinatorial recommendation policy which can handle a large number of candidate items efficiently. In our experiments with real data, we show this generative adversarial user model can better explain user behavior than alternatives, and the RL policy based on this model can lead to a better long-term reward for the user and higher click rate for the system.'
volume: 97
URL: https://proceedings.mlr.press/v97/chen19f.html
PDF: http://proceedings.mlr.press/v97/chen19f/chen19f.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chen19f.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Xinshi
family: Chen
- given: Shuang
family: Li
- given: Hui
family: Li
- given: Shaohua
family: Jiang
- given: Yuan
family: Qi
- given: Le
family: Song
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1052-1061
id: chen19f
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1052
lastpage: 1061
published: 2019-05-24 00:00:00 +0000
- title: 'Understanding and Utilizing Deep Neural Networks Trained with Noisy Labels'
abstract: 'Noisy labels are ubiquitous in real-world datasets, which poses a challenge for robustly training deep neural networks (DNNs) as DNNs usually have the high capacity to memorize the noisy labels. In this paper, we find that the test accuracy can be quantitatively characterized in terms of the noise ratio in datasets. In particular, the test accuracy is a quadratic function of the noise ratio in the case of symmetric noise, which explains the experimental findings previously published. Based on our analysis, we apply cross-validation to randomly split noisy datasets, which identifies most samples that have correct labels. Then we adopt the Co-teaching strategy which takes full advantage of the identified samples to train DNNs robustly against noisy labels. Compared with extensive state-of-the-art methods, our strategy consistently improves the generalization performance of DNNs under both synthetic and real-world training noise.'
volume: 97
URL: https://proceedings.mlr.press/v97/chen19g.html
PDF: http://proceedings.mlr.press/v97/chen19g/chen19g.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chen19g.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Pengfei
family: Chen
- given: Ben Ben
family: Liao
- given: Guangyong
family: Chen
- given: Shengyu
family: Zhang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1062-1070
id: chen19g
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1062
lastpage: 1070
published: 2019-05-24 00:00:00 +0000
- title: 'A Gradual, Semi-Discrete Approach to Generative Network Training via Explicit Wasserstein Minimization'
abstract: 'This paper provides a simple procedure to fit generative networks to target distributions, with the goal of a small Wasserstein distance (or other optimal transport costs). The approach is based on two principles: (a) if the source randomness of the network is a continuous distribution (the "semi-discrete" setting), then the Wasserstein distance is realized by a deterministic optimal transport mapping; (b) given an optimal transport mapping between a generator network and a target distribution, the Wasserstein distance may be decreased via a regression between the generated data and the mapped target points. The procedure here therefore alternates these two steps, forming an optimal transport and regressing against it, gradually adjusting the generator network towards the target distribution. Mathematically, this approach is shown to minimize the Wasserstein distance to both the empirical target distribution, and also its underlying population counterpart. Empirically, good performance is demonstrated on the training and testing sets of the MNIST and Thin-8 data. The paper closes with a discussion of the unsuitability of the Wasserstein distance for certain tasks, as has been identified in prior work (Arora et al., 2017; Huang et al., 2017).'
volume: 97
URL: https://proceedings.mlr.press/v97/chen19h.html
PDF: http://proceedings.mlr.press/v97/chen19h/chen19h.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chen19h.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yucheng
family: Chen
- given: Matus
family: Telgarsky
- given: Chao
family: Zhang
- given: Bolton
family: Bailey
- given: Daniel
family: Hsu
- given: Jian
family: Peng
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1071-1080
id: chen19h
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1071
lastpage: 1080
published: 2019-05-24 00:00:00 +0000
- title: 'Transferability vs. Discriminability: Batch Spectral Penalization for Adversarial Domain Adaptation'
abstract: 'Adversarial domain adaptation has made remarkable advances in learning transferable representations for knowledge transfer across domains. While adversarial learning strengthens the feature transferability which the community focuses on, its impact on the feature discriminability has not been fully explored. In this paper, a series of experiments based on spectral analysis of the feature representations have been conducted, revealing an unexpected deterioration of the discriminability while learning transferable features adversarially. Our key finding is that the eigenvectors with the largest singular values will dominate the feature transferability. As a consequence, the transferability is enhanced at the expense of over penalization of other eigenvectors that embody rich structures crucial for discriminability. Towards this problem, we present Batch Spectral Penalization (BSP), a general approach to penalizing the largest singular values so that other eigenvectors can be relatively strengthened to boost the feature discriminability. Experiments show that the approach significantly improves upon representative adversarial domain adaptation methods to yield state of the art results.'
volume: 97
URL: https://proceedings.mlr.press/v97/chen19i.html
PDF: http://proceedings.mlr.press/v97/chen19i/chen19i.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chen19i.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Xinyang
family: Chen
- given: Sinan
family: Wang
- given: Mingsheng
family: Long
- given: Jianmin
family: Wang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1081-1090
id: chen19i
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1081
lastpage: 1090
published: 2019-05-24 00:00:00 +0000
- title: 'Fast Incremental von Neumann Graph Entropy Computation: Theory, Algorithm, and Applications'
abstract: 'The von Neumann graph entropy (VNGE) facilitates measurement of information divergence and distance between graphs in a graph sequence. It has been successfully applied to various learning tasks driven by network-based data. While effective, VNGE is computationally demanding as it requires the full eigenspectrum of the graph Laplacian matrix. In this paper, we propose a new computational framework, Fast Incremental von Neumann Graph EntRopy (FINGER), which approaches VNGE with a performance guarantee. FINGER reduces the cubic complexity of VNGE to linear complexity in the number of nodes and edges, and thus enables online computation based on incremental graph changes. We also show asymptotic equivalence of FINGER to the exact VNGE, and derive its approximation error bounds. Based on FINGER, we propose efficient algorithms for computing Jensen-Shannon distance between graphs. Our experimental results on different random graph models demonstrate the computational efficiency and the asymptotic equivalence of FINGER. In addition, we apply FINGER to two real-world applications and one synthesized anomaly detection dataset, and corroborate its superior performance over seven baseline graph similarity methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/chen19j.html
PDF: http://proceedings.mlr.press/v97/chen19j/chen19j.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chen19j.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Pin-Yu
family: Chen
- given: Lingfei
family: Wu
- given: Sijia
family: Liu
- given: Indika
family: Rajapakse
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1091-1101
id: chen19j
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1091
lastpage: 1101
published: 2019-05-24 00:00:00 +0000
- title: 'Katalyst: Boosting Convex Katayusha for Non-Convex Problems with a Large Condition Number'
abstract: 'An important class of non-convex objectives that has wide applications in machine learning consists of a sum of $n$ smooth functions and a non-smooth convex function. Tremendous studies have been devoted to conquering these problems by leveraging one of the two types of variance reduction techniques, i.e., SVRG-type that computes a full gradient occasionally and SAGA-type that maintains $n$ stochastic gradients at every iteration. In practice, SVRG-type is preferred to SAGA-type due to its potentially less memory costs. An interesting question that has been largely ignored is how to improve the complexity of variance reduction methods for problems with a large condition number that measures the degree to which the objective is close to a convex function. In this paper, we present a simple but non-trivial boosting of a state-of-the-art SVRG-type method for convex problems (namely Katyusha) to enjoy an improved complexity for solving non-convex problems with a large condition number (that is close to a convex function). To the best of our knowledge, its complexity has the best dependence on $n$ and the degree of non-convexity, and also matches that of a recent SAGA-type accelerated stochastic algorithm for a constrained non-convex smooth optimization problem.'
volume: 97
URL: https://proceedings.mlr.press/v97/chen19k.html
PDF: http://proceedings.mlr.press/v97/chen19k/chen19k.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chen19k.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Zaiyi
family: Chen
- given: Yi
family: Xu
- given: Haoyuan
family: Hu
- given: Tianbao
family: Yang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1102-1111
id: chen19k
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1102
lastpage: 1111
published: 2019-05-24 00:00:00 +0000
- title: 'Multivariate-Information Adversarial Ensemble for Scalable Joint Distribution Matching'
abstract: 'A broad range of cross-$m$-domain generation researches boil down to matching a joint distribution by deep generative models (DGMs). Hitherto algorithms excel in pairwise domains while as $m$ increases, remain struggling to scale themselves to ﬁt a joint distribution. In this paper, we propose a domain-scalable DGM, i.e., MMI-ALI for $m$-domain joint distribution matching. As an $m$-domain ensemble model of ALIs (Dumoulin et al., 2016), MMI-ALI is adversarially trained with maximizing Multivariate Mutual Information (MMI) w.r.t. joint variables of each pair of domains and their shared feature. The negative MMIs are upper bounded by a series of feasible losses provably leading to matching $m$-domain joint distributions. MMI-ALI linearly scales as $m$ increases and thus, strikes a right balance between efﬁcacy and scalability. We evaluate MMI-ALI in diverse challenging $m$-domain scenarios and verify its superiority.'
volume: 97
URL: https://proceedings.mlr.press/v97/chen19l.html
PDF: http://proceedings.mlr.press/v97/chen19l/chen19l.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chen19l.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ziliang
family: Chen
- given: Zhanfu
family: Yang
- given: Xiaoxi
family: Wang
- given: Xiaodan
family: Liang
- given: Xiaopeng
family: Yan
- given: Guanbin
family: Li
- given: Liang
family: Lin
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1112-1121
id: chen19l
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1112
lastpage: 1121
published: 2019-05-24 00:00:00 +0000
- title: 'Robust Decision Trees Against Adversarial Examples'
abstract: 'Although adversarial examples and model robust-ness have been extensively studied in the context of neural networks, research on this issue in tree-based models and how to make tree-based models robust against adversarial examples is still limited. In this paper, we show that tree-based models are also vulnerable to adversarial examples and develop a novel algorithm to learn robust trees. At its core, our method aims to optimize the performance under the worst-case perturbation of input features, which leads to a max-min saddle point problem. Incorporating this saddle point objective into the decision tree building procedure is non-trivial due to the discrete nature of trees{—}a naive approach to finding the best split according to this saddle point objective will take exponential time. To make our approach practical and scalable, we propose efficient tree building algorithms by approximating the inner minimizer in the saddlepoint problem, and present efficient implementations for classical information gain based trees as well as state-of-the-art tree boosting systems such as XGBoost. Experimental results on real world datasets demonstrate that the proposed algorithms can significantly improve the robustness of tree-based models against adversarial examples.'
volume: 97
URL: https://proceedings.mlr.press/v97/chen19m.html
PDF: http://proceedings.mlr.press/v97/chen19m/chen19m.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chen19m.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Hongge
family: Chen
- given: Huan
family: Zhang
- given: Duane
family: Boning
- given: Cho-Jui
family: Hsieh
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1122-1131
id: chen19m
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1122
lastpage: 1131
published: 2019-05-24 00:00:00 +0000
- title: 'RaFM: Rank-Aware Factorization Machines'
abstract: 'Fatorization machines (FM) are a popular model class to learn pairwise interactions by a low-rank approximation. Different from existing FM-based approaches which use a fixed rank for all features, this paper proposes a Rank-Aware FM (RaFM) model which adopts pairwise interactions from embeddings with different ranks. The proposed model achieves a better performance on real-world datasets where different features have significantly varying frequencies of occurrences. Moreover, we prove that the RaFM model can be stored, evaluated, and trained as efficiently as one single FM, and under some reasonable conditions it can be even significantly more efficient than FM. RaFM improves the performance of FMs in both regression tasks and classification tasks while incurring less computational burden, therefore also has attractive potential in industrial applications.'
volume: 97
URL: https://proceedings.mlr.press/v97/chen19n.html
PDF: http://proceedings.mlr.press/v97/chen19n/chen19n.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chen19n.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Xiaoshuang
family: Chen
- given: Yin
family: Zheng
- given: Jiaxing
family: Wang
- given: Wenye
family: Ma
- given: Junzhou
family: Huang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1132-1140
id: chen19n
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1132
lastpage: 1140
published: 2019-05-24 00:00:00 +0000
- title: 'Control Regularization for Reduced Variance Reinforcement Learning'
abstract: 'Dealing with high variance is a significant challenge in model-free reinforcement learning (RL). Existing methods are unreliable, exhibiting high variance in performance from run to run using different initializations/seeds. Focusing on problems arising in continuous control, we propose a functional regularization approach to augmenting model-free RL. In particular, we regularize the behavior of the deep policy to be similar to a policy prior, i.e., we regularize in function space. We show that functional regularization yields a bias-variance trade-off, and propose an adaptive tuning strategy to optimize this trade-off. When the policy prior has control-theoretic stability guarantees, we further show that this regularization approximately preserves those stability guarantees throughout learning. We validate our approach empirically on a range of settings, and demonstrate significantly reduced variance, guaranteed dynamic stability, and more efficient learning than deep RL alone.'
volume: 97
URL: https://proceedings.mlr.press/v97/cheng19a.html
PDF: http://proceedings.mlr.press/v97/cheng19a/cheng19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cheng19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Richard
family: Cheng
- given: Abhinav
family: Verma
- given: Gabor
family: Orosz
- given: Swarat
family: Chaudhuri
- given: Yisong
family: Yue
- given: Joel
family: Burdick
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1141-1150
id: cheng19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1141
lastpage: 1150
published: 2019-05-24 00:00:00 +0000
- title: 'Predictor-Corrector Policy Optimization'
abstract: 'We present a predictor-corrector framework, called PicCoLO, that can transform a first-order model-free reinforcement or imitation learning algorithm into a new hybrid method that leverages predictive models to accelerate policy learning. The new “PicCoLOed” algorithm optimizes a policy by recursively repeating two steps: In the Prediction Step, the learner uses a model to predict the unseen future gradient and then applies the predicted estimate to update the policy; in the Correction Step, the learner runs the updated policy in the environment, receives the true gradient, and then corrects the policy using the gradient error. Unlike previous algorithms, PicCoLO corrects for the mistakes of using imperfect predicted gradients and hence does not suffer from model bias. The development of PicCoLO is made possible by a novel reduction from predictable online learning to adversarial online learning, which provides a systematic way to modify existing first-order algorithms to achieve the optimal regret with respect to predictable information. We show, in both theory and simulation, that the convergence rate of several first-order model-free algorithms can be improved by PicCoLO.'
volume: 97
URL: https://proceedings.mlr.press/v97/cheng19b.html
PDF: http://proceedings.mlr.press/v97/cheng19b/cheng19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cheng19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ching-An
family: Cheng
- given: Xinyan
family: Yan
- given: Nathan
family: Ratliff
- given: Byron
family: Boots
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1151-1161
id: cheng19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1151
lastpage: 1161
published: 2019-05-24 00:00:00 +0000
- title: 'Variational Inference for sparse network reconstruction from count data'
abstract: 'Networks provide a natural yet statistically grounded way to depict and understand how a set of entities interact. However, in many situations interactions are not directly observed and the network needs to be reconstructed based on observations collected for each entity. Our work focuses on the situation where these observations consist of counts. A typical example is the reconstruction of an ecological network based on abundance data. In this setting, the abundance of a set of species is collected in a series of samples and/or environments and we aim at inferring direct interactions between the species. The abundances at hand can be, for example, direct counts of individuals (ecology of macro-organisms) or read counts resulting from metagenomic sequencing (microbial ecology). Whatever the approach chosen to infer such a network, it has to account for the peculiaraties of the data at hand. The first, obvious one, is that the data are counts, i.e. non continuous. Also, the observed counts often vary over many orders of magnitude and are more dispersed than expected under a simple model, such as the Poisson distribution. The observed counts may also result from different sampling efforts in each sample and/or for each entity, which hampers direct comparison. Furthermore, because the network is supposed to reveal only direct interactions, it is highly desirable to account for covariates describing the environment to avoid spurious edges. Many methods of network reconstruction from count data have been proposed. In the context of microbial ecology, most methods (SparCC, REBACCA, SPIEC-EASI, gCODA, BanOCC) rely on a two-step strategy: transform the counts to pseudo Gaussian observations using simple transforms before moving back to the setting of Gaussian Graphical Models, for which state of the art methods exist to infer the network, but only in a Gaussian world. In this work, we consider instead a full-fledged probabilistic model with a latent layer where the counts follow Poisson distributions, conditional to latent (hidden) Gaussian correlated variables. In this model, known as Poisson log-normal (PLN), the dependency structure is completely captured by the latent layer and we model counts, rather than transformations thereof. To our knowledge, the PLN framework is quite new and has only been used by two other recent methods (Mint and plnDAG) to reconstruct networks from count data. In this work, we use the same mathematical framework but adopt a different optimization strategy which alleviates the whole optimization process. We also fully exploit the connection between the PLN framework and generalized linear models to account for the peculiarities of microbiological data sets. The network inference step is done as usual by adding sparsity inducing constraints on the inverse covariance matrix of the latent Gaussian vector to select only the most important interactions between species. Unlike the usual Gaussian setting, the penalized likelihood is generally not tractable in this framework. We resort instead to a variational approximation for parameter inference and solve the corresponding optimization problem by alternating a gradient descent on the variational parameters and a graphical-Lasso step on the covariance matrix. We also select the sparsity parameter using the resampling-based StARS procedure. We show that the sparse PLN approach has better performance than existing methods on simulated datasets and that it extracts relevant signal from microbial ecology datasets. We also show that the inference scales to datasets made up of hundred of species and samples, in line with other methods in the field. In short, our contributions to the field are the following: we extend the use of PLN distributions in network inference by (i) accounting for covariates and offset and thus removing some spurious edges induced by confounding factors, (ii) accounting for different sampling effort to integrate data sets from different sources and thus infer interactions between different types of organisms (e.g. bacteria - fungi), (iii) developing an inference procedure based on the iterative optimization of a well defined objective function. Our objective function is a provable lower bound of the observed likelihood and our procedure accounts for the uncertainty associated with the estimation of the latent variable, unlike the algorithm presented in Mint and plnDAG.'
volume: 97
URL: https://proceedings.mlr.press/v97/chiquet19a.html
PDF: http://proceedings.mlr.press/v97/chiquet19a/chiquet19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chiquet19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Julien
family: Chiquet
- given: Stephane
family: Robin
- given: Mahendra
family: Mariadassou
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1162-1171
id: chiquet19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1162
lastpage: 1171
published: 2019-05-24 00:00:00 +0000
- title: 'Random Walks on Hypergraphs with Edge-Dependent Vertex Weights'
abstract: 'Hypergraphs are used in machine learning to model higher-order relationships in data. While spectral methods for graphs are well-established, spectral theory for hypergraphs remains an active area of research. In this paper, we use random walks to develop a spectral theory for hypergraphs with edge-dependent vertex weights: hypergraphs where every vertex v has a weight $\gamma_e(v)$ for each incident hyperedge e that describes the contribution of v to the hyperedge e. We derive a random walk-based hypergraph Laplacian, and bound the mixing time of random walks on such hypergraphs. Moreover, we give conditions under which random walks on such hypergraphs are equivalent to random walks on graphs. As a corollary, we show that current machine learning methods that rely on Laplacians derived from random walks on hypergraphs with edge-independent vertex weights do not utilize higher-order relationships in the data. Finally, we demonstrate the advantages of hypergraphs with edge-dependent vertex weights on ranking applications using real-world datasets.'
volume: 97
URL: https://proceedings.mlr.press/v97/chitra19a.html
PDF: http://proceedings.mlr.press/v97/chitra19a/chitra19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chitra19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Uthsav
family: Chitra
- given: Benjamin
family: Raphael
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1172-1181
id: chitra19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1172
lastpage: 1181
published: 2019-05-24 00:00:00 +0000
- title: 'Neural Joint Source-Channel Coding'
abstract: 'For reliable transmission across a noisy communication channel, classical results from information theory show that it is asymptotically optimal to separate out the source and channel coding processes. However, this decomposition can fall short in the finite bit-length regime, as it requires non-trivial tuning of hand-crafted codes and assumes infinite computational power for decoding. In this work, we propose to jointly learn the encoding and decoding processes using a new discrete variational autoencoder model. By adding noise into the latent codes to simulate the channel during training, we learn to both compress and error-correct given a fixed bit-length and computational budget. We obtain codes that are not only competitive against several separation schemes, but also learn useful robust representations of the data for downstream tasks such as classification. Finally, inference amortization yields an extremely fast neural decoder, almost an order of magnitude faster compared to standard decoding methods based on iterative belief propagation.'
volume: 97
URL: https://proceedings.mlr.press/v97/choi19a.html
PDF: http://proceedings.mlr.press/v97/choi19a/choi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-choi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Kristy
family: Choi
- given: Kedar
family: Tatwawadi
- given: Aditya
family: Grover
- given: Tsachy
family: Weissman
- given: Stefano
family: Ermon
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1182-1192
id: choi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1182
lastpage: 1192
published: 2019-05-24 00:00:00 +0000
- title: 'Beyond Backprop: Online Alternating Minimization with Auxiliary Variables'
abstract: 'Despite significant recent advances in deep neural networks, training them remains a challenge due to the highly non-convex nature of the objective function. State-of-the-art methods rely on error backpropagation, which suffers from several well-known issues, such as vanishing and exploding gradients, inability to handle non-differentiable nonlinearities and to parallelize weight-updates across layers, and biological implausibility. These limitations continue to motivate exploration of alternative training algorithms, including several recently proposed auxiliary-variable methods which break the complex nested objective function into local subproblems. However, those techniques are mainly offline (batch), which limits their applicability to extremely large datasets, as well as to online, continual or reinforcement learning. The main contribution of our work is a novel online (stochastic/mini-batch) alternating minimization (AM) approach for training deep neural networks, together with the first theoretical convergence guarantees for AM in stochastic settings and promising empirical results on a variety of architectures and datasets.'
volume: 97
URL: https://proceedings.mlr.press/v97/choromanska19a.html
PDF: http://proceedings.mlr.press/v97/choromanska19a/choromanska19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-choromanska19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Anna
family: Choromanska
- given: Benjamin
family: Cowen
- given: Sadhana
family: Kumaravel
- given: Ronny
family: Luss
- given: Mattia
family: Rigotti
- given: Irina
family: Rish
- given: Paolo
family: Diachille
- given: Viatcheslav
family: Gurev
- given: Brian
family: Kingsbury
- given: Ravi
family: Tejwani
- given: Djallel
family: Bouneffouf
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1193-1202
id: choromanska19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1193
lastpage: 1202
published: 2019-05-24 00:00:00 +0000
- title: 'Unifying Orthogonal Monte Carlo Methods'
abstract: 'Many machine learning methods making use of Monte Carlo sampling in vector spaces have been shown to be improved by conditioning samples to be mutually orthogonal. Exact orthogonal coupling of samples is computationally intensive, hence approximate methods have been of great interest. In this paper, we present a unifying perspective of many approximate methods by considering Givens transformations, propose new approximate methods based on this framework, and demonstrate the ﬁrst statistical guarantees for families of approximate methods in kernel approximation. We provide extensive empirical evaluations with guidance for practitioners.'
volume: 97
URL: https://proceedings.mlr.press/v97/choromanski19a.html
PDF: http://proceedings.mlr.press/v97/choromanski19a/choromanski19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-choromanski19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Krzysztof
family: Choromanski
- given: Mark
family: Rowland
- given: Wenyu
family: Chen
- given: Adrian
family: Weller
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1203-1212
id: choromanski19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1203
lastpage: 1212
published: 2019-05-24 00:00:00 +0000
- title: 'Probability Functional Descent: A Unifying Perspective on GANs, Variational Inference, and Reinforcement Learning'
abstract: 'The goal of this paper is to provide a unifying view of a wide range of problems of interest in machine learning by framing them as the minimization of functionals defined on the space of probability measures. In particular, we show that generative adversarial networks, variational inference, and actor-critic methods in reinforcement learning can all be seen through the lens of our framework. We then discuss a generic optimization algorithm for our formulation, called probability functional descent (PFD), and show how this algorithm recovers existing methods developed independently in the settings mentioned earlier.'
volume: 97
URL: https://proceedings.mlr.press/v97/chu19a.html
PDF: http://proceedings.mlr.press/v97/chu19a/chu19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chu19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Casey
family: Chu
- given: Jose
family: Blanchet
- given: Peter
family: Glynn
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1213-1222
id: chu19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1213
lastpage: 1222
published: 2019-05-24 00:00:00 +0000
- title: 'MeanSum: A Neural Model for Unsupervised Multi-Document Abstractive Summarization'
abstract: 'Abstractive summarization has been studied using neural sequence transduction methods with datasets of large, paired document-summary examples. However, such datasets are rare and the models trained from them do not generalize to other domains. Recently, some progress has been made in learning sequence-to-sequence mappings with only unpaired examples. In our work, we consider the setting where there are only documents (product or business reviews) with no summaries provided, and propose an end-to-end, neural model architecture to perform unsupervised abstractive summarization. Our proposed model consists of an auto-encoder where the mean of the representations of the input reviews decodes to a reasonable summary-review. We consider variants of the proposed architecture and perform an ablation study to show the importance of specific components. We show through metrics and human evaluation that the generated summaries are highly abstractive, fluent, relevant, and representative of the average sentiment of the input reviews. Finally, we collect a ground-truth evaluation dataset and show that our model outperforms a strong extractive baseline.'
volume: 97
URL: https://proceedings.mlr.press/v97/chu19b.html
PDF: http://proceedings.mlr.press/v97/chu19b/chu19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chu19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Eric
family: Chu
- given: Peter
family: Liu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1223-1232
id: chu19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1223
lastpage: 1232
published: 2019-05-24 00:00:00 +0000
- title: 'Weak Detection of Signal in the Spiked Wigner Model'
abstract: 'We consider the problem of detecting the presence of the signal in a rank-one signal-plus-noise data matrix. In case the signal-to-noise ratio is under the threshold below which a reliable detection is impossible, we propose a hypothesis test based on the linear spectral statistics of the data matrix. When the noise is Gaussian, the error of the proposed test is optimal as it matches the error of the likelihood ratio test that minimizes the sum of the Type-I and Type-II errors. The test is data-driven and does not depend on the distribution of the signal or the noise. If the density of the noise is known, it can be further improved by an entrywise transformation to lower the error of the test.'
volume: 97
URL: https://proceedings.mlr.press/v97/chung19a.html
PDF: http://proceedings.mlr.press/v97/chung19a/chung19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-chung19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Hye Won
family: Chung
- given: Ji Oon
family: Lee
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1233-1241
id: chung19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1233
lastpage: 1241
published: 2019-05-24 00:00:00 +0000
- title: 'New results on information theoretic clustering'
abstract: 'We study the problem of optimizing the clustering of a set of vectors when the quality of the clustering is measured by the Entropy or the Gini impurity measure. Our results contribute to the state of the art both in terms of best known approximation guarantees and inapproximability bounds: (i) we give the first polynomial time algorithm for Entropy impurity based clustering with approximation guarantee independent of the number of vectors and (ii) we show that the problem of clustering based on entropy impurity does not admit a PTAS. This also implies an inapproximability result in information theoretic clustering for probability distributions closing a problem left open in [Chaudhury and McGregor, COLT08] and [Ackermann et al., ECCC11]. We also report experiments with a new clustering method that was designed on top of the theoretical tools leading to the above results. These experiments suggest a practical applicability for our method, in particular, when the number of clusters is large.'
volume: 97
URL: https://proceedings.mlr.press/v97/cicalese19a.html
PDF: http://proceedings.mlr.press/v97/cicalese19a/cicalese19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cicalese19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ferdinando
family: Cicalese
- given: Eduardo
family: Laber
- given: Lucas
family: Murtinho
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1242-1251
id: cicalese19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1242
lastpage: 1251
published: 2019-05-24 00:00:00 +0000
- title: 'Sensitivity Analysis of Linear Structural Causal Models'
abstract: 'Causal inference requires assumptions about the data generating process, many of which are unverifiable from the data. Given that some causal assumptions might be uncertain or disputed, formal methods are needed to quantify how sensitive research conclusions are to violations of those assumptions. Although an extensive literature exists on the topic, most results are limited to specific model structures, while a general-purpose algorithmic framework for sensitivity analysis is still lacking. In this paper, we develop a formal, systematic approach to sensitivity analysis for arbitrary linear Structural Causal Models (SCMs). We start by formalizing sensitivity analysis as a constrained identification problem. We then develop an efficient, graph-based identification algorithm that exploits non-zero constraints on both directed and bidirected edges. This allows researchers to systematically derive sensitivity curves for a target causal quantity with an arbitrary set of path coefficients and error covariances as sensitivity parameters. These results can be used to display the degree to which violations of causal assumptions affect the target quantity of interest, and to judge, on scientific grounds, whether problematic degrees of violations are plausible.'
volume: 97
URL: https://proceedings.mlr.press/v97/cinelli19a.html
PDF: http://proceedings.mlr.press/v97/cinelli19a/cinelli19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cinelli19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Carlos
family: Cinelli
- given: Daniel
family: Kumor
- given: Bryant
family: Chen
- given: Judea
family: Pearl
- given: Elias
family: Bareinboim
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1252-1261
id: cinelli19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1252
lastpage: 1261
published: 2019-05-24 00:00:00 +0000
- title: 'Dimensionality Reduction for Tukey Regression'
abstract: 'We give the first dimensionality reduction methods for the overconstrained Tukey regression problem. The Tukey loss function $\|y\|_M = \sum_i M(y_i)$ has $M(y_i) \approx |y_i|^p$ for residual errors $y_i$ smaller than a prescribed threshold $\tau$, but $M(y_i)$ becomes constant for errors $|y_i| > \tau$. Our results depend on a new structural result, proven constructively, showing that for any $d$-dimensional subspace $L \subset \mathbb{R}^n$, there is a fixed bounded-size subset of coordinates containing, for every $y \in L$, all the large coordinates, with respect to the Tukey loss function, of $y$. Our methods reduce a given Tukey regression problem to a smaller weighted version, whose solution is a provably good approximate solution to the original problem. Our reductions are fast, simple and easy to implement, and we give empirical results demonstrating their practicality, using existing heuristic solvers for the small versions. We also give exponential-time algorithms giving provably good solutions, and hardness results suggesting that a significant speedup in the worst case is unlikely.'
volume: 97
URL: https://proceedings.mlr.press/v97/clarkson19a.html
PDF: http://proceedings.mlr.press/v97/clarkson19a/clarkson19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-clarkson19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Kenneth
family: Clarkson
- given: Ruosong
family: Wang
- given: David
family: Woodruff
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1262-1271
id: clarkson19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1262
lastpage: 1271
published: 2019-05-24 00:00:00 +0000
- title: 'On Medians of (Randomized) Pairwise Means'
abstract: 'Tournament procedures, recently introduced in the literature, offer an appealing alternative, from a theoretical perspective at least, to the principle of Empirical Risk Minimization in machine learning. Statistical learning by Median-of-Means (MoM) basically consists in segmenting the training data into blocks of equal size and comparing the statistical performance of every pair of candidate decision rules on each data block: that with highest performance on the majority of the blocks is declared as the winner. In the context of nonparametric regression, functions having won all their duels have been shown to outperform empirical risk minimizers w.r.t. the mean squared error under minimal assumptions, while exhibiting robustness properties. It is the purpose of this paper to extend this approach, in order to address other learning problems in particular, for which the performance criterion takes the form of an expectation over pairs of observations rather than over one single observation, as may be the case in pairwise ranking, clustering or metric learning. Precisely, it is proved here that the bounds achieved by MoM are essentially conserved when the blocks are built by means of independent sampling without replacement schemes instead of a simple segmentation. These results are next extended to situations where the risk is related to a pairwise loss function and its empirical counterpart is of the form of a $U$-statistic. Beyond theoretical results guaranteeing the performance of the learning/estimation methods proposed, some numerical experiments provide empirical evidence of their relevance in practice.'
volume: 97
URL: https://proceedings.mlr.press/v97/clemencon19a.html
PDF: http://proceedings.mlr.press/v97/clemencon19a/clemencon19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-clemencon19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Pierre
family: Laforgue
- given: Stephan
family: Clemencon
- given: Patrice
family: Bertail
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1272-1281
id: clemencon19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1272
lastpage: 1281
published: 2019-05-24 00:00:00 +0000
- title: 'Quantifying Generalization in Reinforcement Learning'
abstract: 'In this paper, we investigate the problem of overfitting in deep reinforcement learning. Among the most common benchmarks in RL, it is customary to use the same environments for both training and testing. This practice offers relatively little insight into an agent’s ability to generalize. We address this issue by using procedurally generated environments to construct distinct training and test sets. Most notably, we introduce a new environment called CoinRun, designed as a benchmark for generalization in RL. Using CoinRun, we find that agents overfit to surprisingly large training sets. We then show that deeper convolutional architectures improve generalization, as do methods traditionally found in supervised learning, including L2 regularization, dropout, data augmentation and batch normalization.'
volume: 97
URL: https://proceedings.mlr.press/v97/cobbe19a.html
PDF: http://proceedings.mlr.press/v97/cobbe19a/cobbe19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cobbe19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Karl
family: Cobbe
- given: Oleg
family: Klimov
- given: Chris
family: Hesse
- given: Taehoon
family: Kim
- given: John
family: Schulman
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1282-1289
id: cobbe19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1282
lastpage: 1289
published: 2019-05-24 00:00:00 +0000
- title: 'Empirical Analysis of Beam Search Performance Degradation in Neural Sequence Models'
abstract: 'Beam search is the most popular inference algorithm for decoding neural sequence models. Unlike greedy search, beam search allows for non-greedy local decisions that can potentially lead to a sequence with a higher overall probability. However, work on a number of applications has found that the quality of the highest probability hypothesis found by beam search degrades with large beam widths. We perform an empirical study of the behavior of beam search across three sequence synthesis tasks. We find that increasing the beam width leads to sequences that are disproportionately based on early, very low probability tokens that are followed by a sequence of tokens with higher (conditional) probability. We show that, empirically, such sequences are more likely to have a lower evaluation score than lower probability sequences without this pattern. Using the notion of search discrepancies from heuristic search, we hypothesize that large discrepancies are the cause of the performance degradation. We show that this hypothesis generalizes the previous ones in machine translation and image captioning. To validate our hypothesis, we show that constraining beam search to avoid large discrepancies eliminates the performance degradation.'
volume: 97
URL: https://proceedings.mlr.press/v97/cohen19a.html
PDF: http://proceedings.mlr.press/v97/cohen19a/cohen19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cohen19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Eldan
family: Cohen
- given: Christopher
family: Beck
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1290-1299
id: cohen19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1290
lastpage: 1299
published: 2019-05-24 00:00:00 +0000
- title: 'Learning Linear-Quadratic Regulators Efficiently with only $\sqrtT$ Regret'
abstract: 'We present the first computationally-efficient algorithm with $\widetilde{O}(\sqrt{T})$ regret for learning in Linear Quadratic Control systems with unknown dynamics. By that, we resolve an open question of Abbasi-Yadkori and Szepesvari (2011) and Dean,Mania, Matni, Recht, and Tu (2018).'
volume: 97
URL: https://proceedings.mlr.press/v97/cohen19b.html
PDF: http://proceedings.mlr.press/v97/cohen19b/cohen19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cohen19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alon
family: Cohen
- given: Tomer
family: Koren
- given: Yishay
family: Mansour
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1300-1309
id: cohen19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1300
lastpage: 1309
published: 2019-05-24 00:00:00 +0000
- title: 'Certified Adversarial Robustness via Randomized Smoothing'
abstract: 'We show how to turn any classifier that classifies well under Gaussian noise into a new classifier that is certifiably robust to adversarial perturbations under the L2 norm. While this "randomized smoothing" technique has been proposed before in the literature, we are the first to provide a tight analysis, which establishes a close connection between L2 robustness and Gaussian noise. We use the technique to train an ImageNet classifier with e.g. a certified top-1 accuracy of 49% under adversarial perturbations with L2 norm less than 0.5 (=127/255). Smoothing is the only approach to certifiably robust classification which has been shown feasible on full-resolution ImageNet. On smaller-scale datasets where competing approaches to certified L2 robustness are viable, smoothing delivers higher certified accuracies. The empirical success of the approach suggests that provable methods based on randomization at prediction time are a promising direction for future research into adversarially robust classification.'
volume: 97
URL: https://proceedings.mlr.press/v97/cohen19c.html
PDF: http://proceedings.mlr.press/v97/cohen19c/cohen19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cohen19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jeremy
family: Cohen
- given: Elan
family: Rosenfeld
- given: Zico
family: Kolter
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1310-1320
id: cohen19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1310
lastpage: 1320
published: 2019-05-24 00:00:00 +0000
- title: 'Gauge Equivariant Convolutional Networks and the Icosahedral CNN'
abstract: 'The principle of equivariance to symmetry transformations enables a theoretically grounded approach to neural network architecture design. Equivariant networks have shown excellent performance and data efficiency on vision and medical imaging problems that exhibit symmetries. Here we show how this principle can be extended beyond global symmetries to local gauge transformations. This enables the development of a very general class of convolutional neural networks on manifolds that depend only on the intrinsic geometry, and which includes many popular methods from equivariant and geometric deep learning. We implement gauge equivariant CNNs for signals defined on the surface of the icosahedron, which provides a reasonable approximation of the sphere. By choosing to work with this very regular manifold, we are able to implement the gauge equivariant convolution using a single conv2d call, making it a highly scalable and practical alternative to Spherical CNNs. Using this method, we demonstrate substantial improvements over previous methods on the task of segmenting omnidirectional images and global climate patterns.'
volume: 97
URL: https://proceedings.mlr.press/v97/cohen19d.html
PDF: http://proceedings.mlr.press/v97/cohen19d/cohen19d.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cohen19d.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Taco
family: Cohen
- given: Maurice
family: Weiler
- given: Berkay
family: Kicanaoglu
- given: Max
family: Welling
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1321-1330
id: cohen19d
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1321
lastpage: 1330
published: 2019-05-24 00:00:00 +0000
- title: 'CURIOUS: Intrinsically Motivated Modular Multi-Goal Reinforcement Learning'
abstract: 'In open-ended environments, autonomous learning agents must set their own goals and build their own curriculum through an intrinsically motivated exploration. They may consider a large diversity of goals, aiming to discover what is controllable in their environments, and what is not. Because some goals might prove easy and some impossible, agents must actively select which goal to practice at any moment, to maximize their overall mastery on the set of learnable goals. This paper proposes CURIOUS , an algorithm that leverages 1) a modular Universal Value Function Approximator with hindsight learning to achieve a diversity of goals of different kinds within a unique policy and 2) an automated curriculum learning mechanism that biases the attention of the agent towards goals maximizing the absolute learning progress. Agents focus sequentially on goals of increasing complexity, and focus back on goals that are being forgotten. Experiments conducted in a new modular-goal robotic environment show the resulting developmental self-organization of a learning curriculum, and demonstrate properties of robustness to distracting goals, forgetting and changes in body properties.'
volume: 97
URL: https://proceedings.mlr.press/v97/colas19a.html
PDF: http://proceedings.mlr.press/v97/colas19a/colas19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-colas19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Cédric
family: Colas
- given: Pierre
family: Fournier
- given: Mohamed
family: Chetouani
- given: Olivier
family: Sigaud
- given: Pierre-Yves
family: Oudeyer
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1331-1340
id: colas19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1331
lastpage: 1340
published: 2019-05-24 00:00:00 +0000
- title: 'A fully differentiable beam search decoder'
abstract: 'We introduce a new beam search decoder that is fully differentiable, making it possible to optimize at training time through the inference procedure. Our decoder allows us to combine models which operate at different granularities (e.g. acoustic and language models). It can be used when target sequences are not aligned to input sequences by considering all possible alignments between the two. We demonstrate our approach scales by applying it to speech recognition, jointly training acoustic and word-level language models. The system is end-to-end, with gradients flowing through the whole architecture from the word-level transcriptions. Recent research efforts have shown that deep neural networks with attention-based mechanisms can successfully train an acoustic model from the final transcription, while implicitly learning a language model. Instead, we show that it is possible to discriminatively train an acoustic model jointly with an *explicit* and possibly pre-trained language model.'
volume: 97
URL: https://proceedings.mlr.press/v97/collobert19a.html
PDF: http://proceedings.mlr.press/v97/collobert19a/collobert19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-collobert19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ronan
family: Collobert
- given: Awni
family: Hannun
- given: Gabriel
family: Synnaeve
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1341-1350
id: collobert19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1341
lastpage: 1350
published: 2019-05-24 00:00:00 +0000
- title: 'Scalable Metropolis-Hastings for Exact Bayesian Inference with Large Datasets'
abstract: 'Bayesian inference via standard Markov Chain Monte Carlo (MCMC) methods such as Metropolis-Hastings is too computationally intensive to handle large datasets, since the cost per step usually scales like $O(n)$ in the number of data points $n$. We propose the Scalable Metropolis-Hastings (SMH) kernel that only requires processing on average $O(1)$ or even $O(1/\sqrt{n})$ data points per step. This scheme is based on a combination of factorized acceptance probabilities, procedures for fast simulation of Bernoulli processes, and control variate ideas. Contrary to many MCMC subsampling schemes such as fixed step-size Stochastic Gradient Langevin Dynamics, our approach is exact insofar as the invariant distribution is the true posterior and not an approximation to it. We characterise the performance of our algorithm theoretically, and give realistic and verifiable conditions under which it is geometrically ergodic. This theory is borne out by empirical results that demonstrate overall performance benefits over standard Metropolis-Hastings and various subsampling algorithms.'
volume: 97
URL: https://proceedings.mlr.press/v97/cornish19a.html
PDF: http://proceedings.mlr.press/v97/cornish19a/cornish19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cornish19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Rob
family: Cornish
- given: Paul
family: Vanetti
- given: Alexandre
family: Bouchard-Cote
- given: George
family: Deligiannidis
- given: Arnaud
family: Doucet
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1351-1360
id: cornish19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1351
lastpage: 1360
published: 2019-05-24 00:00:00 +0000
- title: 'Adjustment Criteria for Generalizing Experimental Findings'
abstract: 'Generalizing causal effects from a controlled experiment to settings beyond the particular study population is arguably one of the central tasks found in empirical circles. While a proper design and careful execution of the experiment would support, under mild conditions, the validity of inferences about the population in which the experiment was conducted, two challenges make the extrapolation step to different populations somewhat involved, namely, transportability and sampling selection bias. The former is concerned with disparities in the distributions and causal mechanisms between the domain (i.e., settings, population, environment) where the experiment is conducted and where the inferences are intended; the latter with distortions in the sample’s proportions due to preferential selection of units into the study. In this paper, we investigate the assumptions and machinery necessary for using *covariate adjustment* to correct for the biases generated by both of these problems, and generalize experimental data to infer causal effects in a new domain. We derive complete graphical conditions to determine if a set of covariates is admissible for adjustment in this new setting. Building on the graphical characterization, we develop an efficient algorithm that enumerates all possible admissible sets with poly-time delay guarantee; this can be useful for when some variables are preferred over the others due to different costs or amenability to measurement.'
volume: 97
URL: https://proceedings.mlr.press/v97/correa19a.html
PDF: http://proceedings.mlr.press/v97/correa19a/correa19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-correa19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Juan
family: Correa
- given: Jin
family: Tian
- given: Elias
family: Bareinboim
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1361-1369
id: correa19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1361
lastpage: 1369
published: 2019-05-24 00:00:00 +0000
- title: 'Online Learning with Sleeping Experts and Feedback Graphs'
abstract: 'We consider the scenario of online learning with sleeping experts, where not all experts are available at each round, and analyze the general framework of learning with feedback graphs, where the loss observations associated with each expert are characterized by a graph. A critical assumption in this framework is that the loss observations and the set of sleeping experts at each round are independent. We first extend the classical sleeping experts algorithm of Kleinberg et al. 2008 to the feedback graphs scenario, and prove matching upper and lower bounds for the sleeping regret of the resulting algorithm under the independence assumption. Our main contribution is then to relax this assumption, present a more general notion of sleeping regret, and derive a general algorithm with strong theoretical guarantees. We apply this new framework to the important scenario of online learning with abstention, where a learner can elect to abstain from making a prediction at the price of a certain cost. We empirically validate our algorithm against multiple online abstention algorithms on several real-world datasets, showing substantial performance improvements.'
volume: 97
URL: https://proceedings.mlr.press/v97/cortes19a.html
PDF: http://proceedings.mlr.press/v97/cortes19a/cortes19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cortes19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Corinna
family: Cortes
- given: Giulia
family: Desalvo
- given: Claudio
family: Gentile
- given: Mehryar
family: Mohri
- given: Scott
family: Yang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1370-1378
id: cortes19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1370
lastpage: 1378
published: 2019-05-24 00:00:00 +0000
- title: 'Active Learning with Disagreement Graphs'
abstract: 'We present two novel enhancements of an online importance-weighted active learning algorithm IWAL, using the properties of disagreements among hypotheses. The first enhancement, IWALD, prunes the hypothesis set with a more aggressive strategy based on the disagreement graph. We show that IWAL-D improves the generalization performance and the label complexity of the original IWAL, and quantify the improvement in terms of the disagreement graph coefficient. The second enhancement, IZOOM, further improves IWAL-D by adaptively zooming into the current version space and thus reducing the best-in-class error. We show that IZOOM admits favorable theoretical guarantees with the changing hypothesis set. We report experimental results on multiple datasets and demonstrate that the proposed algorithms achieve better test performances than IWAL given the same amount of labeling budget.'
volume: 97
URL: https://proceedings.mlr.press/v97/cortes19b.html
PDF: http://proceedings.mlr.press/v97/cortes19b/cortes19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cortes19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Corinna
family: Cortes
- given: Giulia
family: Desalvo
- given: Mehryar
family: Mohri
- given: Ningshan
family: Zhang
- given: Claudio
family: Gentile
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1379-1387
id: cortes19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1379
lastpage: 1387
published: 2019-05-24 00:00:00 +0000
- title: 'Shape Constraints for Set Functions'
abstract: 'Set functions predict a label from a permutation-invariant variable-size collection of feature vectors. We propose making set functions more understandable and regularized by capturing domain knowledge through shape constraints. We show how prior work in monotonic constraints can be adapted to set functions, and then propose two new shape constraints designed to generalize the conditioning role of weights in a weighted mean. We show how one can train standard functions and set functions that satisfy these shape constraints with a deep lattice network. We propose a nonlinear estimation strategy we call the semantic feature engine that uses set functions with the proposed shape constraints to estimate labels for compound sparse categorical features. Experiments on real-world data show the achieved accuracy is similar to deep sets or deep neural networks, but provides guarantees on the model behavior, which makes it easier to explain and debug.'
volume: 97
URL: https://proceedings.mlr.press/v97/cotter19a.html
PDF: http://proceedings.mlr.press/v97/cotter19a/cotter19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cotter19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Andrew
family: Cotter
- given: Maya
family: Gupta
- given: Heinrich
family: Jiang
- given: Erez
family: Louidor
- given: James
family: Muller
- given: Tamann
family: Narayan
- given: Serena
family: Wang
- given: Tao
family: Zhu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1388-1396
id: cotter19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1388
lastpage: 1396
published: 2019-05-24 00:00:00 +0000
- title: 'Training Well-Generalizing Classifiers for Fairness Metrics and Other Data-Dependent Constraints'
abstract: 'Classifiers can be trained with data-dependent constraints to satisfy fairness goals, reduce churn, achieve a targeted false positive rate, or other policy goals. We study the generalization performance for such constrained optimization problems, in terms of how well the constraints are satisfied at evaluation time, given that they are satisfied at training time. To improve generalization, we frame the problem as a two-player game where one player optimizes the model parameters on a training dataset, and the other player enforces the constraints on an independent validation dataset. We build on recent work in two-player constrained optimization to show that if one uses this two-dataset approach, then constraint generalization can be significantly improved. As we illustrate experimentally, this approach works not only in theory, but also in practice.'
volume: 97
URL: https://proceedings.mlr.press/v97/cotter19b.html
PDF: http://proceedings.mlr.press/v97/cotter19b/cotter19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cotter19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Andrew
family: Cotter
- given: Maya
family: Gupta
- given: Heinrich
family: Jiang
- given: Nathan
family: Srebro
- given: Karthik
family: Sridharan
- given: Serena
family: Wang
- given: Blake
family: Woodworth
- given: Seungil
family: You
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1397-1405
id: cotter19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1397
lastpage: 1405
published: 2019-05-24 00:00:00 +0000
- title: 'Monge blunts Bayes: Hardness Results for Adversarial Training'
abstract: 'The last few years have seen a staggering number of empirical studies of the robustness of neural networks in a model of adversarial perturbations of their inputs. Most rely on an adversary which carries out local modifications within prescribed balls. None however has so far questioned the broader picture: how to frame a *resource-bounded* adversary so that it can be *severely detrimental* to learning, a non-trivial problem which entails at a minimum the choice of loss and classifiers. We suggest a formal answer for losses that satisfy the minimal statistical requirement of being *proper*. We pin down a simple sufficient property for any given class of adversaries to be detrimental to learning, involving a central measure of “harmfulness” which generalizes the well-known class of integral probability metrics. A key feature of our result is that it holds for *all* proper losses, and for a popular subset of these, the optimisation of this central measure appears to be *independent of the loss*. When classifiers are Lipschitz – a now popular approach in adversarial training –, this optimisation resorts to *optimal transport* to make a low-budget compression of class marginals. Toy experiments reveal a finding recently separately observed: training against a sufficiently budgeted adversary of this kind *improves* generalization.'
volume: 97
URL: https://proceedings.mlr.press/v97/cranko19a.html
PDF: http://proceedings.mlr.press/v97/cranko19a/cranko19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cranko19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Zac
family: Cranko
- given: Aditya
family: Menon
- given: Richard
family: Nock
- given: Cheng Soon
family: Ong
- given: Zhan
family: Shi
- given: Christian
family: Walder
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1406-1415
id: cranko19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1406
lastpage: 1415
published: 2019-05-24 00:00:00 +0000
- title: 'Boosted Density Estimation Remastered'
abstract: 'There has recently been a steady increase in the number iterative approaches to density estimation. However, an accompanying burst of formal convergence guarantees has not followed; all results pay the price of heavy assumptions which are often unrealistic or hard to check. The *Generative Adversarial Network (GAN)* literature — seemingly orthogonal to the aforementioned pursuit — has had the side effect of a renewed interest in variational divergence minimisation (notably $f$-GAN). We show how to combine this latter approach and the classical boosting theory in supervised learning to get the first density estimation algorithm that provably achieves geometric convergence under very weak assumptions. We do so by a trick allowing to combine *classifiers* as the sufficient statistics of an exponential family. Our analysis includes an improved variational characterisation of $f$-GAN.'
volume: 97
URL: https://proceedings.mlr.press/v97/cranko19b.html
PDF: http://proceedings.mlr.press/v97/cranko19b/cranko19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cranko19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Zac
family: Cranko
- given: Richard
family: Nock
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1416-1425
id: cranko19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1416
lastpage: 1425
published: 2019-05-24 00:00:00 +0000
- title: 'Submodular Cost Submodular Cover with an Approximate Oracle'
abstract: 'In this work, we study the Submodular Cost Submodular Cover problem, which is to minimize the submodular cost required to ensure that the submodular benefit function exceeds a given threshold. Existing approximation ratios for the greedy algorithm assume a value oracle to the benefit function. However, access to a value oracle is not a realistic assumption for many applications of this problem, where the benefit function is difficult to compute. We present two incomparable approximation ratios for this problem with an approximate value oracle and demonstrate that the ratios take on empirically relevant values through a case study with the Influence Threshold problem in online social networks.'
volume: 97
URL: https://proceedings.mlr.press/v97/crawford19a.html
PDF: http://proceedings.mlr.press/v97/crawford19a/crawford19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-crawford19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Victoria
family: Crawford
- given: Alan
family: Kuhnle
- given: My
family: Thai
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1426-1435
id: crawford19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1426
lastpage: 1435
published: 2019-05-24 00:00:00 +0000
- title: 'Flexibly Fair Representation Learning by Disentanglement'
abstract: 'We consider the problem of learning representations that achieve group and subgroup fairness with respect to multiple sensitive attributes. Taking inspiration from the disentangled representation learning literature, we propose an algorithm for learning compact representations of datasets that are useful for reconstruction and prediction, but are also *flexibly fair*, meaning they can be easily modified at test time to achieve subgroup demographic parity with respect to multiple sensitive attributes and their conjunctions. We show empirically that the resulting encoder—which does not require the sensitive attributes for inference—allows for the adaptation of a single representation to a variety of fair classification tasks with new target labels and subgroup definitions.'
volume: 97
URL: https://proceedings.mlr.press/v97/creager19a.html
PDF: http://proceedings.mlr.press/v97/creager19a/creager19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-creager19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Elliot
family: Creager
- given: David
family: Madras
- given: Joern-Henrik
family: Jacobsen
- given: Marissa
family: Weis
- given: Kevin
family: Swersky
- given: Toniann
family: Pitassi
- given: Richard
family: Zemel
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1436-1445
id: creager19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1436
lastpage: 1445
published: 2019-05-24 00:00:00 +0000
- title: 'Anytime Online-to-Batch, Optimism and Acceleration'
abstract: 'A standard way to obtain convergence guarantees in stochastic convex optimization is to run an online learning algorithm and then output the average of its iterates: the actual iterates of the online learning algorithm do not come with individual guarantees. We close this gap by introducing a black-box modification to any online learning algorithm whose iterates converge to the optimum in stochastic scenarios. We then consider the case of smooth losses, and show that combining our approach with optimistic online learning algorithms immediately yields a fast convergence rate of $O(L/T^{3/2}+\sigma/\sqrt{T})$ on $L$-smooth problems with $\sigma^2$ variance in the gradients. Finally, we provide a reduction that converts any adaptive online algorithm into one that obtains the optimal accelerated rate of $\tilde O(L/T^2 + \sigma/\sqrt{T})$, while still maintaining $\tilde O(1/\sqrt{T})$ convergence in the non-smooth setting. Importantly, our algorithms adapt to $L$ and $\sigma$ automatically: they do not need to know either to obtain these rates.'
volume: 97
URL: https://proceedings.mlr.press/v97/cutkosky19a.html
PDF: http://proceedings.mlr.press/v97/cutkosky19a/cutkosky19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cutkosky19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ashok
family: Cutkosky
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1446-1454
id: cutkosky19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1446
lastpage: 1454
published: 2019-05-24 00:00:00 +0000
- title: 'Matrix-Free Preconditioning in Online Learning'
abstract: 'We provide an online convex optimization algorithm with regret that interpolates between the regret of an algorithm using an optimal preconditioning matrix and one using a diagonal preconditioning matrix. Our regret bound is never worse than that obtained by diagonal preconditioning, and in certain setting even surpasses that of algorithms with full-matrix preconditioning. Importantly, our algorithm runs in the same time and space complexity as online gradient descent. Along the way we incorporate new techniques that mildly streamline and improve logarithmic factors in prior regret analyses. We conclude by benchmarking our algorithm on synthetic data and deep learning tasks.'
volume: 97
URL: https://proceedings.mlr.press/v97/cutkosky19b.html
PDF: http://proceedings.mlr.press/v97/cutkosky19b/cutkosky19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cutkosky19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ashok
family: Cutkosky
- given: Tamas
family: Sarlos
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1455-1464
id: cutkosky19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1455
lastpage: 1464
published: 2019-05-24 00:00:00 +0000
- title: 'Minimal Achievable Sufficient Statistic Learning'
abstract: 'We introduce Minimal Achievable Sufficient Statistic (MASS) Learning, a machine learning training objective for which the minima are minimal sufficient statistics with respect to a class of functions being optimized over (e.g., deep networks). In deriving MASS Learning, we also introduce Conserved Differential Information (CDI), an information-theoretic quantity that {—} unlike standard mutual information {—} can be usefully applied to deterministically-dependent continuous random variables like the input and output of a deep network. In a series of experiments, we show that deep networks trained with MASS Learning achieve competitive performance on supervised learning, regularization, and uncertainty quantification benchmarks.'
volume: 97
URL: https://proceedings.mlr.press/v97/cvitkovic19a.html
PDF: http://proceedings.mlr.press/v97/cvitkovic19a/cvitkovic19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cvitkovic19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Milan
family: Cvitkovic
- given: Günther
family: Koliander
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1465-1474
id: cvitkovic19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1465
lastpage: 1474
published: 2019-05-24 00:00:00 +0000
- title: 'Open Vocabulary Learning on Source Code with a Graph-Structured Cache'
abstract: 'Machine learning models that take computer program source code as input typically use Natural Language Processing (NLP) techniques. However, a major challenge is that code is written using an open, rapidly changing vocabulary due to, e.g., the coinage of new variable and method names. Reasoning over such a vocabulary is not something for which most NLP methods are designed. We introduce a Graph-Structured Cache to address this problem; this cache contains a node for each new word the model encounters with edges connecting each word to its occurrences in the code. We find that combining this graph-structured cache strategy with recent Graph-Neural-Network-based models for supervised learning on code improves the models’ performance on a code completion task and a variable naming task — with over 100% relative improvement on the latter — at the cost of a moderate increase in computation time.'
volume: 97
URL: https://proceedings.mlr.press/v97/cvitkovic19b.html
PDF: http://proceedings.mlr.press/v97/cvitkovic19b/cvitkovic19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-cvitkovic19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Milan
family: Cvitkovic
- given: Badal
family: Singh
- given: Animashree
family: Anandkumar
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1475-1485
id: cvitkovic19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1475
lastpage: 1485
published: 2019-05-24 00:00:00 +0000
- title: 'The Value Function Polytope in Reinforcement Learning'
abstract: 'We establish geometric and topological properties of the space of value functions in finite state-action Markov decision processes. Our main contribution is the characterization of the nature of its shape: a general polytope (Aigner et al., 2010). To demonstrate this result, we exhibit several properties of the structural relationship between policies and value functions including the line theorem, which shows that the value functions of policies constrained on all but one state describe a line segment. Finally, we use this novel perspective and introduce visualizations to enhance the understanding of the dynamics of reinforcement learning algorithms.'
volume: 97
URL: https://proceedings.mlr.press/v97/dadashi19a.html
PDF: http://proceedings.mlr.press/v97/dadashi19a/dadashi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-dadashi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Robert
family: Dadashi
- given: Adrien Ali
family: Taiga
- given: Nicolas Le
family: Roux
- given: Dale
family: Schuurmans
- given: Marc G.
family: Bellemare
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1486-1495
id: dadashi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1486
lastpage: 1495
published: 2019-05-24 00:00:00 +0000
- title: 'Bayesian Optimization Meets Bayesian Optimal Stopping'
abstract: 'Bayesian optimization (BO) is a popular paradigm for optimizing the hyperparameters of machine learning (ML) models due to its sample efficiency. Many ML models require running an iterative training procedure (e.g., stochastic gradient descent). This motivates the question whether information available during the training process (e.g., validation accuracy after each epoch) can be exploited for improving the epoch efficiency of BO algorithms by early-stopping model training under hyperparameter settings that will end up under-performing and hence eliminating unnecessary training epochs. This paper proposes to unify BO (specifically, Gaussian process-upper confidence bound (GP-UCB)) with Bayesian optimal stopping (BO-BOS) to boost the epoch efficiency of BO. To achieve this, while GP-UCB is sample-efficient in the number of function evaluations, BOS complements it with epoch efficiency for each function evaluation by providing a principled optimal stopping mechanism for early stopping. BO-BOS preserves the (asymptotic) no-regret performance of GP-UCB using our specified choice of BOS parameters that is amenable to an elegant interpretation in terms of the exploration-exploitation trade-off. We empirically evaluate the performance of BO-BOS and demonstrate its generality in hyperparameter optimization of ML models and two other interesting applications.'
volume: 97
URL: https://proceedings.mlr.press/v97/dai19a.html
PDF: http://proceedings.mlr.press/v97/dai19a/dai19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-dai19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Zhongxiang
family: Dai
- given: Haibin
family: Yu
- given: Bryan Kian Hsiang
family: Low
- given: Patrick
family: Jaillet
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1496-1506
id: dai19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1496
lastpage: 1506
published: 2019-05-24 00:00:00 +0000
- title: 'Policy Certificates: Towards Accountable Reinforcement Learning'
abstract: 'The performance of a reinforcement learning algorithm can vary drastically during learning because of exploration. Existing algorithms provide little information about the quality of their current policy before executing it, and thus have limited use in high-stakes applications like healthcare. We address this lack of accountability by proposing that algorithms output policy certificates. These certificates bound the sub-optimality and return of the policy in the next episode, allowing humans to intervene when the certified quality is not satisfactory. We further introduce two new algorithms with certificates and present a new framework for theoretical analysis that guarantees the quality of their policies and certificates. For tabular MDPs, we show that computing certificates can even improve the sample-efficiency of optimism-based exploration. As a result, one of our algorithms is the first to achieve minimax-optimal PAC bounds up to lower-order terms, and this algorithm also matches (and in some settings slightly improves upon) existing minimax regret bounds.'
volume: 97
URL: https://proceedings.mlr.press/v97/dann19a.html
PDF: http://proceedings.mlr.press/v97/dann19a/dann19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-dann19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Christoph
family: Dann
- given: Lihong
family: Li
- given: Wei
family: Wei
- given: Emma
family: Brunskill
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1507-1516
id: dann19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1507
lastpage: 1516
published: 2019-05-24 00:00:00 +0000
- title: 'Learning Fast Algorithms for Linear Transforms Using Butterfly Factorizations'
abstract: 'Fast linear transforms are ubiquitous in machine learning, including the discrete Fourier transform, discrete cosine transform, and other structured transformations such as convolutions. All of these transforms can be represented by dense matrix-vector multiplication, yet each has a specialized and highly efficient (subquadratic) algorithm. We ask to what extent hand-crafting these algorithms and implementations is necessary, what structural prior they encode, and how much knowledge is required to automatically learn a fast algorithm for a provided structured transform. Motivated by a characterization of fast matrix-vector multiplication as products of sparse matrices, we introduce a parameterization of divide-and-conquer methods that is capable of representing a large class of transforms. This generic formulation can automatically learn an efficient algorithm for many important transforms; for example, it recovers the $O(N \log N)$ Cooley-Tukey FFT algorithm to machine precision, for dimensions $N$ up to $1024$. Furthermore, our method can be incorporated as a lightweight replacement of generic matrices in machine learning pipelines to learn efficient and compressible transformations. On a standard task of compressing a single hidden-layer network, our method exceeds the classification accuracy of unconstrained matrices on CIFAR-10 by 3.9 points—the first time a structured approach has done so—with 4X faster inference speed and 40X fewer parameters.'
volume: 97
URL: https://proceedings.mlr.press/v97/dao19a.html
PDF: http://proceedings.mlr.press/v97/dao19a/dao19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-dao19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Tri
family: Dao
- given: Albert
family: Gu
- given: Matthew
family: Eichhorn
- given: Atri
family: Rudra
- given: Christopher
family: Re
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1517-1527
id: dao19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1517
lastpage: 1527
published: 2019-05-24 00:00:00 +0000
- title: 'A Kernel Theory of Modern Data Augmentation'
abstract: 'Data augmentation, a technique in which a training set is expanded with class-preserving transformations, is ubiquitous in modern machine learning pipelines. In this paper, we seek to establish a theoretical framework for understanding data augmentation. We approach this from two directions: First, we provide a general model of augmentation as a Markov process, and show that kernels appear naturally with respect to this model, even when we do not employ kernel classification. Next, we analyze more directly the effect of augmentation on kernel classifiers, showing that data augmentation can be approximated by first-order feature averaging and second-order variance regularization components. These frameworks both serve to illustrate the ways in which data augmentation affects the downstream learning model, and the resulting analyses provide novel connections between prior work in invariant kernels, tangent propagation, and robust optimization. Finally, we provide several proof-of-concept applications showing that our theory can be useful for accelerating machine learning workflows, such as reducing the amount of computation needed to train using augmented data, and predicting the utility of a transformation prior to training.'
volume: 97
URL: https://proceedings.mlr.press/v97/dao19b.html
PDF: http://proceedings.mlr.press/v97/dao19b/dao19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-dao19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Tri
family: Dao
- given: Albert
family: Gu
- given: Alexander
family: Ratner
- given: Virginia
family: Smith
- given: Chris
family: De Sa
- given: Christopher
family: Re
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1528-1537
id: dao19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1528
lastpage: 1537
published: 2019-05-24 00:00:00 +0000
- title: 'TarMAC: Targeted Multi-Agent Communication'
abstract: 'We propose a targeted communication architecture for multi-agent reinforcement learning, where agents learn both *what* messages to send and *whom* to address them to while performing cooperative tasks in partially-observable environments. This targeting behavior is learnt solely from downstream task-specific reward without any communication supervision. We additionally augment this with a multi-round communication approach where agents coordinate via multiple rounds of communication before taking actions in the environment. We evaluate our approach on a diverse set of cooperative multi-agent tasks, of varying difficulties, with varying number of agents, in a variety of environments ranging from 2D grid layouts of shapes and simulated traffic junctions to 3D indoor environments, and demonstrate the benefits of targeted and multi-round communication. Moreover, we show that the targeted communication strategies learned by agents are interpretable and intuitive. Finally, we show that our architecture can be easily extended to mixed and competitive environments, leading to improved performance and sample complexity over recent state-of-the-art approaches.'
volume: 97
URL: https://proceedings.mlr.press/v97/das19a.html
PDF: http://proceedings.mlr.press/v97/das19a/das19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-das19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Abhishek
family: Das
- given: Théophile
family: Gervet
- given: Joshua
family: Romoff
- given: Dhruv
family: Batra
- given: Devi
family: Parikh
- given: Mike
family: Rabbat
- given: Joelle
family: Pineau
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1538-1546
id: das19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1538
lastpage: 1546
published: 2019-05-24 00:00:00 +0000
- title: 'Teaching a black-box learner'
abstract: 'One widely-studied model of *teaching* calls for a teacher to provide the minimal set of labeled examples that uniquely specifies a target concept. The assumption is that the teacher knows the learner’s hypothesis class, which is often not true of real-life teaching scenarios. We consider the problem of teaching a learner whose representation and hypothesis class are unknown—that is, the learner is a black box. We show that a teacher who does not interact with the learner can do no better than providing random examples. We then prove, however, that with interaction, a teacher can efficiently find a set of teaching examples that is a provably good approximation to the optimal set. As an illustration, we show how this scheme can be used to *shrink* training sets for any family of classifiers: that is, to find an approximately-minimal subset of training instances that yields the same classifier as the entire set.'
volume: 97
URL: https://proceedings.mlr.press/v97/dasgupta19a.html
PDF: http://proceedings.mlr.press/v97/dasgupta19a/dasgupta19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-dasgupta19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sanjoy
family: Dasgupta
- given: Daniel
family: Hsu
- given: Stefanos
family: Poulis
- given: Xiaojin
family: Zhu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1547-1555
id: dasgupta19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1547
lastpage: 1555
published: 2019-05-24 00:00:00 +0000
- title: 'Stochastic Deep Networks'
abstract: 'Machine learning is increasingly targeting areas where input data cannot be accurately described by a single vector, but can be modeled instead using the more flexible concept of random vectors, namely probability measures or more simply point clouds of varying cardinality. Using deep architectures on measures poses, however, many challenging issues. Indeed, deep architectures are originally designed to handle fixed-length vectors, or, using recursive mechanisms, ordered sequences thereof. In sharp contrast, measures describe a varying number of weighted observations with no particular order. We propose in this work a deep framework designed to handle crucial aspects of measures, namely permutation invariances, variations in weights and cardinality. Architectures derived from this pipeline can (i) map measures to measures - using the concept of push-forward operators; (ii) bridge the gap between measures and Euclidean spaces - through integration steps. This allows to design discriminative networks (to classify or reduce the dimensionality of input measures), generative architectures (to synthesize measures) and recurrent pipelines (to predict measure dynamics). We provide a theoretical analysis of these building blocks, review our architectures’ approximation abilities and robustness w.r.t. perturbation, and try them on various discriminative and generative tasks.'
volume: 97
URL: https://proceedings.mlr.press/v97/de-bie19a.html
PDF: http://proceedings.mlr.press/v97/de-bie19a/de-bie19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-de-bie19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Gwendoline
family: De Bie
- given: Gabriel
family: Peyré
- given: Marco
family: Cuturi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1556-1565
id: de-bie19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1556
lastpage: 1565
published: 2019-05-24 00:00:00 +0000
- title: 'Learning-to-Learn Stochastic Gradient Descent with Biased Regularization'
abstract: 'We study the problem of learning-to-learn: infer- ring a learning algorithm that works well on a family of tasks sampled from an unknown distribution. As class of algorithms we consider Stochastic Gradient Descent (SGD) on the true risk regularized by the square euclidean distance from a bias vector. We present an average excess risk bound for such a learning algorithm that quantifies the potential benefit of using a bias vector with respect to the unbiased case. We then propose a novel meta-algorithm to estimate the bias term online from a sequence of observed tasks. The small memory footprint and low time complexity of our approach makes it appealing in practice while our theoretical analysis provides guarantees on the generalization properties of the meta-algorithm on new tasks. A key feature of our results is that, when the number of tasks grows and their vari- ance is relatively small, our learning-to-learn approach has a significant advantage over learning each task in isolation by standard SGD without a bias term. Numerical experiments demonstrate the effectiveness of our approach in practice.'
volume: 97
URL: https://proceedings.mlr.press/v97/denevi19a.html
PDF: http://proceedings.mlr.press/v97/denevi19a/denevi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-denevi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Giulia
family: Denevi
- given: Carlo
family: Ciliberto
- given: Riccardo
family: Grazzi
- given: Massimiliano
family: Pontil
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1566-1575
id: denevi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1566
lastpage: 1575
published: 2019-05-24 00:00:00 +0000
- title: 'A Multitask Multiple Kernel Learning Algorithm for Survival Analysis with Application to Cancer Biology'
abstract: 'Predictive performance of machine learning algorithms on related problems can be improved using multitask learning approaches. Rather than performing survival analysis on each data set to predict survival times of cancer patients, we developed a novel multitask approach based on multiple kernel learning (MKL). Our multitask MKL algorithm both works on multiple cancer data sets and integrates cancer-related pathways/gene sets into survival analysis. We tested our algorithm, which is named as Path2MSurv, on the Cancer Genome Atlas data sets analyzing gene expression profiles of 7,655 patients from 20 cancer types together with cancer-specific pathway/gene set collections. Path2MSurv obtained better or comparable predictive performance when benchmarked against random survival forest, survival support vector machine, and single-task variant of our algorithm. Path2MSurv has the ability to identify key pathways/gene sets in predicting survival times of patients from different cancer types.'
volume: 97
URL: https://proceedings.mlr.press/v97/dereli19a.html
PDF: http://proceedings.mlr.press/v97/dereli19a/dereli19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-dereli19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Onur
family: Dereli
- given: Ceyda
family: Oğuz
- given: Mehmet
family: Gönen
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1576-1585
id: dereli19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1576
lastpage: 1585
published: 2019-05-24 00:00:00 +0000
- title: 'Learning to Convolve: A Generalized Weight-Tying Approach'
abstract: 'Recent work (Cohen & Welling, 2016) has shown that generalizations of convolutions, based on group theory, provide powerful inductive biases for learning. In these generalizations, filters are not only translated but can also be rotated, flipped, etc. However, coming up with exact models of how to rotate a 3x3 filter on a square pixel-grid is difficult. In this paper, we learn how to transform filters for use in the group convolution, focussing on roto-translation. For this, we learn a filter basis and all rotated versions of that filter basis. Filters are then encoded by a set of rotation invariant coefficients. To rotate a filter, we switch the basis. We demonstrate we can produce feature maps with low sensitivity to input rotations, while achieving high performance on MNIST and CIFAR-10.'
volume: 97
URL: https://proceedings.mlr.press/v97/diaconu19a.html
PDF: http://proceedings.mlr.press/v97/diaconu19a/diaconu19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-diaconu19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Nichita
family: Diaconu
- given: Daniel
family: Worrall
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1586-1595
id: diaconu19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1586
lastpage: 1595
published: 2019-05-24 00:00:00 +0000
- title: 'Sever: A Robust Meta-Algorithm for Stochastic Optimization'
abstract: 'In high dimensions, most machine learning methods are brittle to even a small fraction of structured outliers. To address this, we introduce a new meta-algorithm that can take in a base learner such as least squares or stochastic gradient descent, and harden the learner to be resistant to outliers. Our method, Sever, possesses strong theoretical guarantees yet is also highly scalable – beyond running the base learner itself, it only requires computing the top singular vector of a certain n{\texttimes}d matrix. We apply Sever on a drug design dataset and a spam classification dataset, and find that in both cases it has substantially greater robustness than several baselines. On the spam dataset, with 1% corruptions, we achieved 7.4% test error, compared to 13.4%-20.5% for the baselines, and 3% error on the uncorrupted dataset. Similarly, on the drug design dataset, with 10% corruptions, we achieved 1.42 mean-squared error test error, compared to 1.51-2.33 for the baselines, and 1.23 error on the uncorrupted dataset.'
volume: 97
URL: https://proceedings.mlr.press/v97/diakonikolas19a.html
PDF: http://proceedings.mlr.press/v97/diakonikolas19a/diakonikolas19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-diakonikolas19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ilias
family: Diakonikolas
- given: Gautam
family: Kamath
- given: Daniel
family: Kane
- given: Jerry
family: Li
- given: Jacob
family: Steinhardt
- given: Alistair
family: Stewart
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1596-1606
id: diakonikolas19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1596
lastpage: 1606
published: 2019-05-24 00:00:00 +0000
- title: 'Approximated Oracle Filter Pruning for Destructive CNN Width Optimization'
abstract: 'It is not easy to design and run Convolutional Neural Networks (CNNs) due to: 1) finding the optimal number of filters (i.e., the width) at each layer is tricky, given an architecture; and 2) the computational intensity of CNNs impedes the deployment on computationally limited devices. Oracle Pruning is designed to remove the unimportant filters from a well-trained CNN, which estimates the filters’ importance by ablating them in turn and evaluating the model, thus delivers high accuracy but suffers from intolerable time complexity, and requires a given resulting width but cannot automatically find it. To address these problems, we propose Approximated Oracle Filter Pruning (AOFP), which keeps searching for the least important filters in a binary search manner, makes pruning attempts by masking out filters randomly, accumulates the resulting errors, and finetunes the model via a multi-path framework. As AOFP enables simultaneous pruning on multiple layers, we can prune an existing very deep CNN with acceptable time cost, negligible accuracy drop, and no heuristic knowledge, or re-design a model which exerts higher accuracy and faster inference.'
volume: 97
URL: https://proceedings.mlr.press/v97/ding19a.html
PDF: http://proceedings.mlr.press/v97/ding19a/ding19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ding19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Xiaohan
family: Ding
- given: Guiguang
family: Ding
- given: Yuchen
family: Guo
- given: Jungong
family: Han
- given: Chenggang
family: Yan
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1607-1616
id: ding19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1607
lastpage: 1616
published: 2019-05-24 00:00:00 +0000
- title: 'Noisy Dual Principal Component Pursuit'
abstract: 'Dual Principal Component Pursuit (DPCP) is a recently proposed non-convex optimization based method for learning subspaces of high relative dimension from noiseless datasets contaminated by as many outliers as the square of the number of inliers. Experimentally, DPCP has proved to be robust to noise and outperform the popular RANSAC on 3D vision tasks such as road plane detection and relative poses estimation from three views. This paper extends the global optimality and convergence theory of DPCP to the case of data corrupted by noise, and further demonstrates its robustness using synthetic and real data.'
volume: 97
URL: https://proceedings.mlr.press/v97/ding19b.html
PDF: http://proceedings.mlr.press/v97/ding19b/ding19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ding19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Tianyu
family: Ding
- given: Zhihui
family: Zhu
- given: Tianjiao
family: Ding
- given: Yunchen
family: Yang
- given: Rene
family: Vidal
- given: Manolis
family: Tsakiris
- given: Daniel
family: Robinson
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1617-1625
id: ding19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1617
lastpage: 1625
published: 2019-05-24 00:00:00 +0000
- title: 'Finite-Time Analysis of Distributed TD(0) with Linear Function Approximation on Multi-Agent Reinforcement Learning'
abstract: 'We study the policy evaluation problem in multi-agent reinforcement learning. In this problem, a group of agents works cooperatively to evaluate the value function for the global discounted accumulative reward problem, which is composed of local rewards observed by the agents. Over a series of time steps, the agents act, get rewarded, update their local estimate of the value function, then communicate with their neighbors. The local update at each agent can be interpreted as a distributed consensus-based variant of the popular temporal difference learning algorithm TD(0). While distributed reinforcement learning algorithms have been presented in the literature, almost nothing is known about their convergence rate. Our main contribution is providing a finite-time analysis for the convergence of the distributed TD(0) algorithm. We do this when the communication network between the agents is time-varying in general. We obtain an explicit upper bound on the rate of convergence of this algorithm as a function of the network topology and the discount factor. Our results mirror what we would expect from using distributed stochastic gradient descent for solving convex optimization problems.'
volume: 97
URL: https://proceedings.mlr.press/v97/doan19a.html
PDF: http://proceedings.mlr.press/v97/doan19a/doan19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-doan19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Thinh
family: Doan
- given: Siva
family: Maguluri
- given: Justin
family: Romberg
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1626-1635
id: doan19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1626
lastpage: 1635
published: 2019-05-24 00:00:00 +0000
- title: 'Trajectory-Based Off-Policy Deep Reinforcement Learning'
abstract: 'Policy gradient methods are powerful reinforcement learning algorithms and have been demonstrated to solve many complex tasks. However, these methods are also data-inefficient, afflicted with high variance gradient estimates, and frequently get stuck in local optima. This work addresses these weaknesses by combining recent improvements in the reuse of off-policy data and exploration in parameter space with deterministic behavioral policies. The resulting objective is amenable to standard neural network optimization strategies like stochastic gradient descent or stochastic gradient Hamiltonian Monte Carlo. Incorporation of previous rollouts via importance sampling greatly improves data-efficiency, whilst stochastic optimization schemes facilitate the escape from local optima. We evaluate the proposed approach on a series of continuous control benchmark tasks. The results show that the proposed algorithm is able to successfully and reliably learn solutions using fewer system interactions than standard policy gradient methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/doerr19a.html
PDF: http://proceedings.mlr.press/v97/doerr19a/doerr19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-doerr19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Andreas
family: Doerr
- given: Michael
family: Volpp
- given: Marc
family: Toussaint
- given: Trimpe
family: Sebastian
- given: Christian
family: Daniel
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1636-1645
id: doerr19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1636
lastpage: 1645
published: 2019-05-24 00:00:00 +0000
- title: 'Generalized No Free Lunch Theorem for Adversarial Robustness'
abstract: 'This manuscript presents some new impossibility results on adversarial robustness in machine learning, a very important yet largely open problem. We show that if conditioned on a class label the data distribution satisfies the $W_2$ Talagrand transportation-cost inequality (for example, this condition is satisfied if the conditional distribution has density which is log-concave; is the uniform measure on a compact Riemannian manifold with positive Ricci curvature, any classifier can be adversarially fooled with high probability once the perturbations are slightly greater than the natural noise level in the problem. We call this result The Strong "No Free Lunch" Theorem as some recent results (Tsipras et al. 2018, Fawzi et al. 2018, etc.) on the subject can be immediately recovered as very particular cases. Our theoretical bounds are demonstrated on both simulated and real data (MNIST). We conclude the manuscript with some speculation on possible future research directions.'
volume: 97
URL: https://proceedings.mlr.press/v97/dohmatob19a.html
PDF: http://proceedings.mlr.press/v97/dohmatob19a/dohmatob19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-dohmatob19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Elvis
family: Dohmatob
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1646-1654
id: dohmatob19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1646
lastpage: 1654
published: 2019-05-24 00:00:00 +0000
- title: 'Width Provably Matters in Optimization for Deep Linear Neural Networks'
abstract: 'We prove that for an $L$-layer fully-connected linear neural network, if the width of every hidden layer is $\widetilde{\Omega}\left(L \cdot r \cdot d_{out} \cdot \kappa^3 \right)$, where $r$ and $\kappa$ are the rank and the condition number of the input data, and $d_{out}$ is the output dimension, then gradient descent with Gaussian random initialization converges to a global minimum at a linear rate. The number of iterations to find an $\epsilon$-suboptimal solution is $O(\kappa \log(\frac{1}{\epsilon}))$. Our polynomial upper bound on the total running time for wide deep linear networks and the $\exp\left(\Omega\left(L\right)\right)$ lower bound for narrow deep linear neural networks [Shamir, 2018] together demonstrate that wide layers are necessary for optimizing deep models.'
volume: 97
URL: https://proceedings.mlr.press/v97/du19a.html
PDF: http://proceedings.mlr.press/v97/du19a/du19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-du19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Simon
family: Du
- given: Wei
family: Hu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1655-1664
id: du19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1655
lastpage: 1664
published: 2019-05-24 00:00:00 +0000
- title: 'Provably efficient RL with Rich Observations via Latent State Decoding'
abstract: 'We study the exploration problem in episodic MDPs with rich observations generated from a small number of latent states. Under certain identifiability assumptions, we demonstrate how to estimate a mapping from the observations to latent states inductively through a sequence of regression and clustering steps—where previously decoded latent states provide labels for later regression problems—and use it to construct good exploration policies. We provide finite-sample guarantees on the quality of the learned state decoding function and exploration policies, and complement our theory with an empirical evaluation on a class of hard exploration problems. Our method exponentially improves over $Q$-learning with naïve exploration, even when $Q$-learning has cheating access to latent states.'
volume: 97
URL: https://proceedings.mlr.press/v97/du19b.html
PDF: http://proceedings.mlr.press/v97/du19b/du19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-du19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Simon
family: Du
- given: Akshay
family: Krishnamurthy
- given: Nan
family: Jiang
- given: Alekh
family: Agarwal
- given: Miroslav
family: Dudik
- given: John
family: Langford
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1665-1674
id: du19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1665
lastpage: 1674
published: 2019-05-24 00:00:00 +0000
- title: 'Gradient Descent Finds Global Minima of Deep Neural Networks'
abstract: 'Gradient descent finds a global minimum in training deep neural networks despite the objective function being non-convex. The current paper proves gradient descent achieves zero training loss in polynomial time for a deep over-parameterized neural network with residual connections (ResNet). Our analysis relies on the particular structure of the Gram matrix induced by the neural network architecture. This structure allows us to show the Gram matrix is stable throughout the training process and this stability implies the global optimality of the gradient descent algorithm. We further extend our analysis to deep residual convolutional neural networks and obtain a similar convergence result.'
volume: 97
URL: https://proceedings.mlr.press/v97/du19c.html
PDF: http://proceedings.mlr.press/v97/du19c/du19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-du19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Simon
family: Du
- given: Jason
family: Lee
- given: Haochuan
family: Li
- given: Liwei
family: Wang
- given: Xiyu
family: Zhai
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1675-1685
id: du19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1675
lastpage: 1685
published: 2019-05-24 00:00:00 +0000
- title: 'Incorporating Grouping Information into Bayesian Decision Tree Ensembles'
abstract: 'We consider the problem of nonparametric regression in the high-dimensional setting in which $P \gg N$. We study the use of overlapping group structures to improve prediction and variable selection. These structures arise commonly when analyzing DNA microarray data, where genes can naturally be grouped according to genetic pathways. We incorporate overlapping group structure into a Bayesian additive regression trees model using a prior constructed so that, if a variable from some group is used to construct a split, this increases the probability that subsequent splits will use predictors from the same group. We refer to our model as an overlapping group Bayesian additive regression trees (OG-BART) model, and our prior on the splits an overlapping group Dirichlet (OG-Dirichlet) prior. Like the sparse group lasso, our prior encourages sparsity both within and between groups. We study the correlation structure of the prior, illustrate the proposed methodology on simulated data, and apply the methodology to gene expression data to learn which genetic pathways are predictive of breast cancer tumor metastasis.'
volume: 97
URL: https://proceedings.mlr.press/v97/du19d.html
PDF: http://proceedings.mlr.press/v97/du19d/du19d.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-du19d.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Junliang
family: Du
- given: Antonio
family: Linero
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1686-1695
id: du19d
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1686
lastpage: 1695
published: 2019-05-24 00:00:00 +0000
- title: 'Task-Agnostic Dynamics Priors for Deep Reinforcement Learning'
abstract: 'While model-based deep reinforcement learning (RL) holds great promise for sample efficiency and generalization, learning an accurate dynamics model is often challenging and requires substantial interaction with the environment. A wide variety of domains have dynamics that share common foundations like the laws of classical mechanics, which are rarely exploited by existing algorithms. In fact, humans continuously acquire and use such dynamics priors to easily adapt to operating in new environments. In this work, we propose an approach to learn task-agnostic dynamics priors from videos and incorporate them into an RL agent. Our method involves pre-training a frame predictor on task-agnostic physics videos to initialize dynamics models (and fine-tune them) for unseen target environments. Our frame prediction architecture, SpatialNet, is designed specifically to capture localized physical phenomena and interactions. Our approach allows for both faster policy learning and convergence to better policies, outperforming competitive approaches on several different environments. We also demonstrate that incorporating this prior allows for more effective transfer between environments.'
volume: 97
URL: https://proceedings.mlr.press/v97/du19e.html
PDF: http://proceedings.mlr.press/v97/du19e/du19e.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-du19e.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yilun
family: Du
- given: Karthic
family: Narasimhan
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1696-1705
id: du19e
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1696
lastpage: 1705
published: 2019-05-24 00:00:00 +0000
- title: 'Optimal Auctions through Deep Learning'
abstract: 'Designing an incentive compatible auction that maximizes expected revenue is an intricate task. The single-item case was resolved in a seminal piece of work by Myerson in 1981. Even after 30-40 years of intense research the problem remains unsolved for seemingly simple multi-bidder, multi-item settings. In this work, we initiate the exploration of the use of tools from deep learning for the automated design of optimal auctions. We model an auction as a multi-layer neural network, frame optimal auction design as a constrained learning problem, and show how it can be solved using standard pipelines. We prove generalization bounds and present extensive experiments, recovering essentially all known analytical solutions for multi-item settings, and obtaining novel mechanisms for settings in which the optimal mechanism is unknown.'
volume: 97
URL: https://proceedings.mlr.press/v97/duetting19a.html
PDF: http://proceedings.mlr.press/v97/duetting19a/duetting19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-duetting19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Paul
family: Duetting
- given: Zhe
family: Feng
- given: Harikrishna
family: Narasimhan
- given: David
family: Parkes
- given: Sai Srivatsa
family: Ravindranath
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1706-1715
id: duetting19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1706
lastpage: 1715
published: 2019-05-24 00:00:00 +0000
- title: 'Wasserstein of Wasserstein Loss for Learning Generative Models'
abstract: 'The Wasserstein distance serves as a loss function for unsupervised learning which depends on the choice of a ground metric on sample space. We propose to use the Wasserstein distance itself as the ground metric on the sample space of images. This ground metric is known as an effective distance for image retrieval, that correlates with human perception. We derive the Wasserstein ground metric on pixel space and define a Riemannian Wasserstein gradient penalty to be used in the Wasserstein Generative Adversarial Network (WGAN) framework. The new gradient penalty is computed efficiently via convolutions on the $L^2$ gradients with negligible additional computational cost. The new formulation is more robust to the natural variability of the data and provides for a more continuous discriminator in sample space.'
volume: 97
URL: https://proceedings.mlr.press/v97/dukler19a.html
PDF: http://proceedings.mlr.press/v97/dukler19a/dukler19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-dukler19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yonatan
family: Dukler
- given: Wuchen
family: Li
- given: Alex
family: Lin
- given: Guido
family: Montufar
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1716-1725
id: dukler19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1716
lastpage: 1725
published: 2019-05-24 00:00:00 +0000
- title: 'Learning interpretable continuous-time models of latent stochastic dynamical systems'
abstract: 'We develop an approach to learn an interpretable semi-parametric model of a latent continuous-time stochastic dynamical system, assuming noisy high-dimensional outputs sampled at uneven times. The dynamics are described by a nonlinear stochastic differential equation (SDE) driven by a Wiener process, with a drift evolution function drawn from a Gaussian process (GP) conditioned on a set of learnt fixed points and corresponding local Jacobian matrices. This form yields a flexible nonparametric model of the dynamics, with a representation corresponding directly to the interpretable portraits routinely employed in the study of nonlinear dynamical systems. The learning algorithm combines inference of continuous latent paths underlying observed data with a sparse variational description of the dynamical process. We demonstrate our approach on simulated data from different nonlinear dynamical systems.'
volume: 97
URL: https://proceedings.mlr.press/v97/duncker19a.html
PDF: http://proceedings.mlr.press/v97/duncker19a/duncker19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-duncker19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Lea
family: Duncker
- given: Gergo
family: Bohner
- given: Julien
family: Boussard
- given: Maneesh
family: Sahani
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1726-1734
id: duncker19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1726
lastpage: 1734
published: 2019-05-24 00:00:00 +0000
- title: 'Autoregressive Energy Machines'
abstract: 'Neural density estimators are flexible families of parametric models which have seen widespread use in unsupervised machine learning in recent years. Maximum-likelihood training typically dictates that these models be constrained to specify an explicit density. However, this limitation can be overcome by instead using a neural network to specify an energy function, or unnormalized density, which can subsequently be normalized to obtain a valid distribution. The challenge with this approach lies in accurately estimating the normalizing constant of the high-dimensional energy function. We propose the Autoregressive Energy Machine, an energy-based model which simultaneously learns an unnormalized density and computes an importance-sampling estimate of the normalizing constant for each conditional in an autoregressive decomposition. The Autoregressive Energy Machine achieves state-of-the-art performance on a suite of density-estimation tasks.'
volume: 97
URL: https://proceedings.mlr.press/v97/durkan19a.html
PDF: http://proceedings.mlr.press/v97/durkan19a/durkan19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-durkan19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Charlie
family: Nash
- given: Conor
family: Durkan
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1735-1744
id: durkan19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1735
lastpage: 1744
published: 2019-05-24 00:00:00 +0000
- title: 'Band-limited Training and Inference for Convolutional Neural Networks'
abstract: 'The convolutional layers are core building blocks of neural network architectures. In general, a convolutional filter applies to the entire frequency spectrum of the input data. We explore artificially constraining the frequency spectra of these filters and data, called band-limiting, during training. The frequency domain constraints apply to both the feed-forward and back-propagation steps. Experimentally, we observe that Convolutional Neural Networks (CNNs) are resilient to this compression scheme and results suggest that CNNs learn to leverage lower-frequency components. In particular, we found: (1) band-limited training can effectively control the resource usage (GPU and memory); (2) models trained with band-limited layers retain high prediction accuracy; and (3) requires no modification to existing training algorithms or neural network architectures to use unlike other compression schemes.'
volume: 97
URL: https://proceedings.mlr.press/v97/dziedzic19a.html
PDF: http://proceedings.mlr.press/v97/dziedzic19a/dziedzic19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-dziedzic19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Adam
family: Dziedzic
- given: John
family: Paparrizos
- given: Sanjay
family: Krishnan
- given: Aaron
family: Elmore
- given: Michael
family: Franklin
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1745-1754
id: dziedzic19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1745
lastpage: 1754
published: 2019-05-24 00:00:00 +0000
- title: 'Imitating Latent Policies from Observation'
abstract: 'In this paper, we describe a novel approach to imitation learning that infers latent policies directly from state observations. We introduce a method that characterizes the causal effects of latent actions on observations while simultaneously predicting their likelihood. We then outline an action alignment procedure that leverages a small amount of environment interactions to determine a mapping between the latent and real-world actions. We show that this corrected labeling can be used for imitating the observed behavior, even though no expert actions are given. We evaluate our approach within classic control environments and a platform game and demonstrate that it performs better than standard approaches. Code for this work is available at https://github.com/ashedwards/ILPO.'
volume: 97
URL: https://proceedings.mlr.press/v97/edwards19a.html
PDF: http://proceedings.mlr.press/v97/edwards19a/edwards19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-edwards19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ashley
family: Edwards
- given: Himanshu
family: Sahni
- given: Yannick
family: Schroecker
- given: Charles
family: Isbell
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1755-1763
id: edwards19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1755
lastpage: 1763
published: 2019-05-24 00:00:00 +0000
- title: 'Semi-Cyclic Stochastic Gradient Descent'
abstract: 'We consider convex SGD updates with a block-cyclic structure, i.e., where each cycle consists of a small number of blocks, each with many samples from a possibly different, block-specific, distribution. This situation arises, e.g., in Federated Learning where the mobile devices available for updates at different times during the day have different characteristics. We show that such block-cyclic structure can significantly deteriorate the performance of SGD, but propose a simple approach that allows prediction with the same guarantees as for i.i.d., non-cyclic, sampling.'
volume: 97
URL: https://proceedings.mlr.press/v97/eichner19a.html
PDF: http://proceedings.mlr.press/v97/eichner19a/eichner19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-eichner19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Hubert
family: Eichner
- given: Tomer
family: Koren
- given: Brendan
family: Mcmahan
- given: Nathan
family: Srebro
- given: Kunal
family: Talwar
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1764-1773
id: eichner19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1764
lastpage: 1773
published: 2019-05-24 00:00:00 +0000
- title: 'GDPP: Learning Diverse Generations using Determinantal Point Processes'
abstract: 'Generative models have proven to be an outstanding tool for representing high-dimensional probability distributions and generating realistic looking images. An essential characteristic of generative models is their ability to produce multi-modal outputs. However, while training, they are often susceptible to mode collapse, that is models are limited in mapping input noise to only a few modes of the true data distribution. In this work, we draw inspiration from Determinantal Point Process (DPP) to propose an unsupervised penalty loss that alleviates mode collapse while producing higher quality samples. DPP is an elegant probabilistic measure used to model negative correlations within a subset and hence quantify its diversity. We use DPP kernel to model the diversity in real data as well as in synthetic data. Then, we devise an objective term that encourages generator to synthesize data with a similar diversity to real data. In contrast to previous state-of-the-art generative models that tend to use additional trainable parameters or complex training paradigms, our method does not change the original training scheme. Embedded in an adversarial training and variational autoencoder, our Generative DPP approach shows a consistent resistance to mode-collapse on a wide-variety of synthetic data and natural image datasets including MNIST, CIFAR10, and CelebA, while outperforming state-of-the-art methods for data-efficiency, generation quality, and convergence-time whereas being 5.8x faster than its closest competitor.'
volume: 97
URL: https://proceedings.mlr.press/v97/elfeki19a.html
PDF: http://proceedings.mlr.press/v97/elfeki19a/elfeki19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-elfeki19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Mohamed
family: Elfeki
- given: Camille
family: Couprie
- given: Morgane
family: Riviere
- given: Mohamed
family: Elhoseiny
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1774-1783
id: elfeki19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1774
lastpage: 1783
published: 2019-05-24 00:00:00 +0000
- title: 'Sequential Facility Location: Approximate Submodularity and Greedy Algorithm'
abstract: 'We develop and analyze a novel utility function and a fast optimization algorithm for subset selection in sequential data that incorporates the dynamic model of data. We propose a cardinality-constrained sequential facility location function that finds a fixed number of representatives, where the sequence of representatives is compatible with the dynamic model and well encodes the data. As maximizing this new objective function is NP-hard, we develop a fast greedy algorithm based on submodular maximization. Unlike the conventional facility location, the computation of the marginal gain in our case cannot be done by operations on each item independently. We exploit the sequential structure of the problem and develop an efficient dynamic programming-based algorithm that computes the marginal gain exactly. We investigate conditions on the dynamic model, under which our utility function is ($\epsilon$-approximately) submodualr, hence, the greedy algorithm comes with performance guarantees. By experiments on synthetic data and the problem of procedure learning from instructional videos, we show that our framework significantly improves the computational time, achieves better objective function values and obtains more coherent summaries.'
volume: 97
URL: https://proceedings.mlr.press/v97/elhamifar19a.html
PDF: http://proceedings.mlr.press/v97/elhamifar19a/elhamifar19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-elhamifar19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ehsan
family: Elhamifar
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1784-1793
id: elhamifar19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1784
lastpage: 1793
published: 2019-05-24 00:00:00 +0000
- title: 'Improved Convergence for $\ell_1$ and $\ell_∞$ Regression via Iteratively Reweighted Least Squares'
abstract: 'The iteratively reweighted least squares method (IRLS) is a popular technique used in practice for solving regression problems. Various versions of this method have been proposed, but their theoretical analyses failed to capture the good practical performance. In this paper we propose a simple and natural version of IRLS for solving $\ell_\infty$ and $\ell_1$ regression, which provably converges to a $(1+\epsilon)$-approximate solution in $O(m^{1/3}\log(1/\epsilon)/\epsilon^{2/3} + \log m/\epsilon^2)$ iterations, where $m$ is the number of rows of the input matrix. Interestingly, this running time is independent of the conditioning of the input, and the dominant term of the running time depends sublinearly in $\epsilon^{-1}$, which is atypical for the optimization of non-smooth functions. This improves upon the more complex algorithms of Chin et al. (ITCS ’12), and Christiano et al. (STOC ’11) by a factor of at least $1/\epsilon^2$, and yields a truly efficient natural algorithm for the slime mold dynamics (Straszak-Vishnoi, SODA ’16, ITCS ’16, ITCS ’17).'
volume: 97
URL: https://proceedings.mlr.press/v97/ene19a.html
PDF: http://proceedings.mlr.press/v97/ene19a/ene19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ene19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alina
family: Ene
- given: Adrian
family: Vladu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1794-1801
id: ene19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1794
lastpage: 1801
published: 2019-05-24 00:00:00 +0000
- title: 'Exploring the Landscape of Spatial Robustness'
abstract: 'The study of adversarial robustness has so far largely focused on perturbations bound in $\ell_p$-norms. However, state-of-the-art models turn out to be also vulnerable to other, more natural classes of perturbations such as translations and rotations. In this work, we thoroughly investigate the vulnerability of neural network–based classifiers to rotations and translations. While data augmentation offers relatively small robustness, we use ideas from robust optimization and test-time input aggregation to significantly improve robustness. Finally we find that, in contrast to the $\ell_p$-norm case, first-order methods cannot reliably find worst-case perturbations. This highlights spatial robustness as a fundamentally different setting requiring additional study.'
volume: 97
URL: https://proceedings.mlr.press/v97/engstrom19a.html
PDF: http://proceedings.mlr.press/v97/engstrom19a/engstrom19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-engstrom19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Logan
family: Engstrom
- given: Brandon
family: Tran
- given: Dimitris
family: Tsipras
- given: Ludwig
family: Schmidt
- given: Aleksander
family: Madry
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1802-1811
id: engstrom19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1802
lastpage: 1811
published: 2019-05-24 00:00:00 +0000
- title: 'Cross-Domain 3D Equivariant Image Embeddings'
abstract: 'Spherical convolutional networks have been introduced recently as tools to learn powerful feature representations of 3D shapes. Spherical CNNs are equivariant to 3D rotations making them ideally suited to applications where 3D data may be observed in arbitrary orientations. In this paper we learn 2D image embeddings with a similar equivariant structure: embedding the image of a 3D object should commute with rotations of the object. We introduce a cross-domain embedding from 2D images into a spherical CNN latent space. This embedding encodes images with 3D shape properties and is equivariant to 3D rotations of the observed object. The model is supervised only by target embeddings obtained from a spherical CNN pretrained for 3D shape classification. We show that learning a rich embedding for images with appropriate geometric structure is sufficient for tackling varied applications, such as relative pose estimation and novel view synthesis, without requiring additional task-specific supervision.'
volume: 97
URL: https://proceedings.mlr.press/v97/esteves19a.html
PDF: http://proceedings.mlr.press/v97/esteves19a/esteves19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-esteves19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Carlos
family: Esteves
- given: Avneesh
family: Sud
- given: Zhengyi
family: Luo
- given: Kostas
family: Daniilidis
- given: Ameesh
family: Makadia
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1812-1822
id: esteves19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1812
lastpage: 1822
published: 2019-05-24 00:00:00 +0000
- title: 'On the Connection Between Adversarial Robustness and Saliency Map Interpretability'
abstract: 'Recent studies on the adversarial vulnerability of neural networks have shown that models trained to be more robust to adversarial attacks exhibit more interpretable saliency maps than their non-robust counterparts. We aim to quantify this behaviour by considering the alignment between input image and saliency map. We hypothesize that as the distance to the decision boundary grows, so does the alignment. This connection is strictly true in the case of linear models. We confirm these theoretical findings with experiments based on models trained with a local Lipschitz regularization and identify where the nonlinear nature of neural networks weakens the relation.'
volume: 97
URL: https://proceedings.mlr.press/v97/etmann19a.html
PDF: http://proceedings.mlr.press/v97/etmann19a/etmann19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-etmann19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Christian
family: Etmann
- given: Sebastian
family: Lunz
- given: Peter
family: Maass
- given: Carola
family: Schoenlieb
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1823-1832
id: etmann19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1823
lastpage: 1832
published: 2019-05-24 00:00:00 +0000
- title: 'Non-monotone Submodular Maximization with Nearly Optimal Adaptivity and Query Complexity'
abstract: 'Submodular maximization is a general optimization problem with a wide range of applications in machine learning (e.g., active learning, clustering, and feature selection). In large-scale optimization, the parallel running time of an algorithm is governed by its adaptivity, which measures the number of sequential rounds needed if the algorithm can execute polynomially-many independent oracle queries in parallel. While low adaptivity is ideal, it is not sufficient for an algorithm to be efficient in practice—there are many applications of distributed submodular optimization where the number of function evaluations becomes prohibitively expensive. Motivated by these applications, we study the adaptivity and query complexity of submodular maximization. In this paper, we give the first constant-factor approximation algorithm for maximizing a non-monotone submodular function subject to a cardinality constraint $k$ that runs in $O(\log(n))$ adaptive rounds and makes $O(n \log(k))$ oracle queries in expectation. In our empirical study, we use three real-world applications to compare our algorithm with several benchmarks for non-monotone submodular maximization. The results demonstrate that our algorithm finds competitive solutions using significantly fewer rounds and queries.'
volume: 97
URL: https://proceedings.mlr.press/v97/fahrbach19a.html
PDF: http://proceedings.mlr.press/v97/fahrbach19a/fahrbach19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-fahrbach19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Matthew
family: Fahrbach
- given: Vahab
family: Mirrokni
- given: Morteza
family: Zadimoghaddam
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1833-1842
id: fahrbach19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1833
lastpage: 1842
published: 2019-05-24 00:00:00 +0000
- title: 'Multi-Frequency Vector Diffusion Maps'
abstract: 'We introduce multi-frequency vector diffusion maps (MFVDM), a new framework for organizing and analyzing high dimensional data sets. The new method is a mathematical and algorithmic generalization of vector diffusion maps (VDM) and other non-linear dimensionality reduction methods. The idea of MFVDM is to incorporates multiple unitary irreducible representations of the alignment group which introduces robustness to noise. We illustrate the efficacy of MFVDM on synthetic and cryo-EM image datasets, achieving better nearest neighbors search and alignment estimation than other baselines as VDM and diffusion maps (DM), especially on extremely noisy data.'
volume: 97
URL: https://proceedings.mlr.press/v97/fan19a.html
PDF: http://proceedings.mlr.press/v97/fan19a/fan19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-fan19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yifeng
family: Fan
- given: Zhizhen
family: Zhao
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1843-1852
id: fan19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1843
lastpage: 1852
published: 2019-05-24 00:00:00 +0000
- title: 'Stable-Predictive Optimistic Counterfactual Regret Minimization'
abstract: 'The CFR framework has been a powerful tool for solving large-scale extensive-form games in practice. However, the theoretical rate at which past CFR-based algorithms converge to the Nash equilibrium is on the order of $O(T^{-1/2})$, where $T$ is the number of iterations. In contrast, first-order methods can be used to achieve a $O(T^{-1})$ dependence on iterations, yet these methods have been less successful in practice. In this work we present the first CFR variant that breaks the square-root dependence on iterations. By combining and extending recent advances on predictive and stable regret minimizers for the matrix-game setting we show that it is possible to leverage “optimistic” regret minimizers to achieve a $O(T^{-3/4})$ convergence rate within CFR. This is achieved by introducing a new notion of stable-predictivity, and by setting the stability of each counterfactual regret minimizer relative to its location in the decision tree. Experiments show that this method is faster than the original CFR algorithm, although not as fast as newer variants, in spite of their worst-case $O(T^{-1/2})$ dependence on iterations.'
volume: 97
URL: https://proceedings.mlr.press/v97/farina19a.html
PDF: http://proceedings.mlr.press/v97/farina19a/farina19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-farina19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Gabriele
family: Farina
- given: Christian
family: Kroer
- given: Noam
family: Brown
- given: Tuomas
family: Sandholm
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1853-1862
id: farina19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1853
lastpage: 1862
published: 2019-05-24 00:00:00 +0000
- title: 'Regret Circuits: Composability of Regret Minimizers'
abstract: 'Regret minimization is a powerful tool for solving large-scale problems; it was recently used in breakthrough results for large-scale extensive-form game solving. This was achieved by composing simplex regret minimizers into an overall regret-minimization framework for extensive-form game strategy spaces. In this paper we study the general composability of regret minimizers. We derive a calculus for constructing regret minimizers for composite convex sets that are obtained from convexity-preserving operations on simpler convex sets. We show that local regret minimizers for the simpler sets can be combined with additional regret minimizers into an aggregate regret minimizer for the composite set. As one application, we show that the CFR framework can be constructed easily from our framework. We also show ways to include curtailing (constraining) operations into our framework. For one, they enable the construction of CFR generalization for extensive-form games with general convex strategy constraints that can cut across decision points.'
volume: 97
URL: https://proceedings.mlr.press/v97/farina19b.html
PDF: http://proceedings.mlr.press/v97/farina19b/farina19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-farina19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Gabriele
family: Farina
- given: Christian
family: Kroer
- given: Tuomas
family: Sandholm
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1863-1872
id: farina19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1863
lastpage: 1872
published: 2019-05-24 00:00:00 +0000
- title: 'Dead-ends and Secure Exploration in Reinforcement Learning'
abstract: 'Many interesting applications of reinforcement learning (RL) involve MDPs that include numerous “dead-end" states. Upon reaching a dead-end state, the agent continues to interact with the environment in a dead-end trajectory before reaching an undesired terminal state, regardless of whatever actions are chosen. The situation is even worse when existence of many dead-end states is coupled with distant positive rewards from any initial state (we term this as Bridge Effect). Hence, conventional exploration techniques often incur prohibitively many training steps before convergence. To deal with the bridge effect, we propose a condition for exploration, called security. We next establish formal results that translate the security condition into the learning problem of an auxiliary value function. This new value function is used to cap “any" given exploration policy and is guaranteed to make it secure. As a special case, we use this theory and introduce secure random-walk. We next extend our results to the deep RL settings by identifying and addressing two main challenges that arise. Finally, we empirically compare secure random-walk with standard benchmarks in two sets of experiments including the Atari game of Montezuma’s Revenge.'
volume: 97
URL: https://proceedings.mlr.press/v97/fatemi19a.html
PDF: http://proceedings.mlr.press/v97/fatemi19a/fatemi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-fatemi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Mehdi
family: Fatemi
- given: Shikhar
family: Sharma
- given: Harm
family: Van Seijen
- given: Samira Ebrahimi
family: Kahou
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1873-1881
id: fatemi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1873
lastpage: 1881
published: 2019-05-24 00:00:00 +0000
- title: 'Invariant-Equivariant Representation Learning for Multi-Class Data'
abstract: 'Representations learnt through deep neural networks tend to be highly informative, but opaque in terms of what information they learn to encode. We introduce an approach to probabilistic modelling that learns to represent data with two separate deep representations: an invariant representation that encodes the information of the class from which the data belongs, and an equivariant representation that encodes the symmetry transformation defining the particular data point within the class manifold (equivariant in the sense that the representation varies naturally with symmetry transformations). This approach is based primarily on the strategic routing of data through the two latent variables, and thus is conceptually transparent, easy to implement, and in-principle generally applicable to any data comprised of discrete classes of continuous distributions (e.g. objects in images, topics in language, individuals in behavioural data). We demonstrate qualitatively compelling representation learning and competitive quantitative performance, in both supervised and semi-supervised settings, versus comparable modelling approaches in the literature with little fine tuning.'
volume: 97
URL: https://proceedings.mlr.press/v97/feige19a.html
PDF: http://proceedings.mlr.press/v97/feige19a/feige19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-feige19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ilya
family: Feige
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1882-1891
id: feige19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1882
lastpage: 1891
published: 2019-05-24 00:00:00 +0000
- title: 'The advantages of multiple classes for reducing overfitting from test set reuse'
abstract: 'Excessive reuse of holdout data can lead to overfitting. However, there is little concrete evidence of significant overfitting due to holdout reuse in popular multiclass benchmarks today. Known results show that, in the worst-case, revealing the accuracy of $k$ adaptively chosen classifiers on a data set of size $n$ allows to create a classifier with bias of $\Theta(\sqrt{k/n})$ for any binary prediction problem. We show a new upper bound of $\tilde O(\max\{\sqrt{k\log(n)/(mn)}, k/n\})$ on the worst-case bias that any attack can achieve in a prediction problem with $m$ classes. Moreover, we present an efficient attack that achieve a bias of $\Omega(\sqrt{k/(m^2 n)})$ and improves on previous work for the binary setting ($m=2$). We also present an inefficient attack that achieves a bias of $\tilde\Omega(k/n)$. Complementing our theoretical work, we give new practical attacks to stress-test multiclass benchmarks by aiming to create as large a bias as possible with a given number of queries. Our experiments show that the additional uncertainty of prediction with a large number of classes indeed mitigates the effect of our best attacks.'
volume: 97
URL: https://proceedings.mlr.press/v97/feldman19a.html
PDF: http://proceedings.mlr.press/v97/feldman19a/feldman19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-feldman19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Vitaly
family: Feldman
- given: Roy
family: Frostig
- given: Moritz
family: Hardt
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1892-1900
id: feldman19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1892
lastpage: 1900
published: 2019-05-24 00:00:00 +0000
- title: 'Decentralized Exploration in Multi-Armed Bandits'
abstract: 'We consider the decentralized exploration problem: a set of players collaborate to identify the best arm by asynchronously interacting with the same stochastic environment. The objective is to insure privacy in the best arm identification problem between asynchronous, collaborative, and thrifty players. In the context of a digital service, we advocate that this decentralized approach allows a good balance between conflicting interests: the providers optimize their services, while protecting privacy of users and saving resources. We define the privacy level as the amount of information an adversary could infer by intercepting all the messages concerning a single user. We provide a generic algorithm DECENTRALIZED ELIMINATION, which uses any best arm identification algorithm as a subroutine. We prove that this algorithm insures privacy, with a low communication cost, and that in comparison to the lower bound of the best arm identification problem, its sample complexity suffers from a penalty depending on the inverse of the probability of the most frequent players. Then, thanks to the genericity of the approach, we extend the proposed algorithm to the non-stationary bandits. Finally, experiments illustrate and complete the analysis.'
volume: 97
URL: https://proceedings.mlr.press/v97/feraud19a.html
PDF: http://proceedings.mlr.press/v97/feraud19a/feraud19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-feraud19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Raphael
family: Feraud
- given: Reda
family: Alami
- given: Romain
family: Laroche
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1901-1909
id: feraud19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1901
lastpage: 1909
published: 2019-05-24 00:00:00 +0000
- title: 'Almost surely constrained convex optimization'
abstract: 'We propose a stochastic gradient framework for solving stochastic composite convex optimization problems with (possibly) infinite number of linear inclusion constraints that need to be satisfied almost surely. We use smoothing and homotopy techniques to handle constraints without the need for matrix-valued projections. We show for our stochastic gradient algorithm $\mathcal{O}(\log(k)/\sqrt{k})$ convergence rate for general convex objectives and $\mathcal{O}(\log(k)/k)$ convergence rate for restricted strongly convex objectives. These rates are known to be optimal up to logarithmic factor, even without constraints. We conduct numerical experiments on basis pursuit, hard margin support vector machines and portfolio optimization problems and show that our algorithm achieves state-of-the-art practical performance.'
volume: 97
URL: https://proceedings.mlr.press/v97/fercoq19a.html
PDF: http://proceedings.mlr.press/v97/fercoq19a/fercoq19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-fercoq19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Olivier
family: Fercoq
- given: Ahmet
family: Alacaoglu
- given: Ion
family: Necoara
- given: Volkan
family: Cevher
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1910-1919
id: fercoq19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1910
lastpage: 1919
published: 2019-05-24 00:00:00 +0000
- title: 'Online Meta-Learning'
abstract: 'A central capability of intelligent systems is the ability to continuously build upon previous experiences to speed up and enhance learning of new tasks. Two distinct research paradigms have studied this question. Meta-learning views this problem as learning a prior over model parameters that is amenable for fast adaptation on a new task, but typically assumes the tasks are available together as a batch. In contrast, online (regret based) learning considers a setting where tasks are revealed one after the other, but conventionally trains a single model without task-specific adaptation. This work introduces an online meta-learning setting, which merges ideas from both paradigms to better capture the spirit and practice of continual lifelong learning. We propose the follow the meta leader (FTML) algorithm which extends the MAML algorithm to this setting. Theoretically, this work provides an O(log T) regret guarantee with one additional higher order smoothness assumption (in comparison to the standard online setting). Our experimental evaluation on three different large-scale problems suggest that the proposed algorithm significantly outperforms alternatives based on traditional online learning approaches.'
volume: 97
URL: https://proceedings.mlr.press/v97/finn19a.html
PDF: http://proceedings.mlr.press/v97/finn19a/finn19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-finn19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Chelsea
family: Finn
- given: Aravind
family: Rajeswaran
- given: Sham
family: Kakade
- given: Sergey
family: Levine
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1920-1930
id: finn19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1920
lastpage: 1930
published: 2019-05-24 00:00:00 +0000
- title: 'DL2: Training and Querying Neural Networks with Logic'
abstract: 'We present DL2, a system for training and querying neural networks with logical constraints. Using DL2, one can declaratively specify domain knowledge constraints to be enforced during training, as well as pose queries on the model to find inputs that satisfy a set of constraints. DL2 works by translating logical constraints into a loss function with desirable mathematical properties. The loss is then minimized with standard gradient-based methods. We evaluate DL2 by training networks with interesting constraints in unsupervised, semi-supervised and supervised settings. Our experimental evaluation demonstrates that DL2 is more expressive than prior approaches combining logic and neural networks, and its loss functions are better suited for optimization. Further, we show that for a number of queries, DL2 can find the desired inputs in seconds (even for large models such as ResNet-50 on ImageNet).'
volume: 97
URL: https://proceedings.mlr.press/v97/fischer19a.html
PDF: http://proceedings.mlr.press/v97/fischer19a/fischer19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-fischer19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Marc
family: Fischer
- given: Mislav
family: Balunovic
- given: Dana
family: Drachsler-Cohen
- given: Timon
family: Gehr
- given: Ce
family: Zhang
- given: Martin
family: Vechev
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1931-1941
id: fischer19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1931
lastpage: 1941
published: 2019-05-24 00:00:00 +0000
- title: 'Bayesian Action Decoder for Deep Multi-Agent Reinforcement Learning'
abstract: 'When observing the actions of others, humans make inferences about why they acted as they did, and what this implies about the world; humans also use the fact that their actions will be interpreted in this manner, allowing them to act informatively and thereby communicate efficiently with others. Although learning algorithms have recently achieved superhuman performance in a number of two-player, zero-sum games, scalable multi-agent reinforcement learning algorithms that can discover effective strategies and conventions in complex, partially observable settings have proven elusive. We present the *Bayesian action decoder* (BAD), a new multi-agent learning method that uses an approximate Bayesian update to obtain a public belief that conditions on the actions taken by all agents in the environment. BAD introduces a new Markov decision process, the *public belief MDP*, in which the action space consists of all deterministic partial policies, and exploits the fact that an agent acting only on this public belief state can still learn to use its private information if the action space is augmented to be over all partial policies mapping private information into environment actions. The Bayesian update is closely related to the *theory of mind* reasoning that humans carry out when observing others’ actions. We first validate BAD on a proof-of-principle two-step matrix game, where it outperforms policy gradient methods; we then evaluate BAD on the challenging, cooperative partial-information card game Hanabi, where, in the two-player setting, it surpasses all previously published learning and hand-coded approaches, establishing a new state of the art.'
volume: 97
URL: https://proceedings.mlr.press/v97/foerster19a.html
PDF: http://proceedings.mlr.press/v97/foerster19a/foerster19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-foerster19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jakob
family: Foerster
- given: Francis
family: Song
- given: Edward
family: Hughes
- given: Neil
family: Burch
- given: Iain
family: Dunning
- given: Shimon
family: Whiteson
- given: Matthew
family: Botvinick
- given: Michael
family: Bowling
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1942-1951
id: foerster19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1942
lastpage: 1951
published: 2019-05-24 00:00:00 +0000
- title: 'Scalable Nonparametric Sampling from Multimodal Posteriors with the Posterior Bootstrap'
abstract: 'Increasingly complex datasets pose a number of challenges for Bayesian inference. Conventional posterior sampling based on Markov chain Monte Carlo can be too computationally intensive, is serial in nature and mixes poorly between posterior modes. Furthermore, all models are misspecified, which brings into question the validity of the conventional Bayesian update. We present a scalable Bayesian nonparametric learning routine that enables posterior sampling through the optimization of suitably randomized objective functions. A Dirichlet process prior on the unknown data distribution accounts for model misspecification, and admits an embarrassingly parallel posterior bootstrap algorithm that generates independent and exact samples from the nonparametric posterior distribution. Our method is particularly adept at sampling from multimodal posterior distributions via a random restart mechanism, and we demonstrate this on Gaussian mixture model and sparse logistic regression examples.'
volume: 97
URL: https://proceedings.mlr.press/v97/fong19a.html
PDF: http://proceedings.mlr.press/v97/fong19a/fong19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-fong19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Edwin
family: Fong
- given: Simon
family: Lyddon
- given: Chris
family: Holmes
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1952-1962
id: fong19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1952
lastpage: 1962
published: 2019-05-24 00:00:00 +0000
- title: 'On discriminative learning of prediction uncertainty'
abstract: 'In classification with a reject option, the classifier is allowed in uncertain cases to abstain from prediction. The classical cost based model of an optimal classifier with a reject option requires the cost of rejection to be defined explicitly. An alternative bounded-improvement model, avoiding the notion of the reject cost, seeks for a classifier with a guaranteed selective risk and maximal cover. We prove that both models share the same class of optimal strategies, and we provide an explicit relation between the reject cost and the target risk being the parameters of the two models. An optimal rejection strategy for both models is based on thresholding the conditional risk defined by posterior probabilities which are usually unavailable. We propose a discriminative algorithm learning an uncertainty function which preserves ordering of the input space induced by the conditional risk, and hence can be used to construct optimal rejection strategies.'
volume: 97
URL: https://proceedings.mlr.press/v97/franc19a.html
PDF: http://proceedings.mlr.press/v97/franc19a/franc19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-franc19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Vojtech
family: Franc
- given: Daniel
family: Prusa
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1963-1971
id: franc19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1963
lastpage: 1971
published: 2019-05-24 00:00:00 +0000
- title: 'Learning Discrete Structures for Graph Neural Networks'
abstract: 'Graph neural networks (GNNs) are a popular class of machine learning models that have been successfully applied to a range of problems. Their major advantage lies in their ability to explicitly incorporate a sparse and discrete dependency structure between data points. Unfortunately, GNNs can only be used when such a graph-structure is available. In practice, however, real-world graphs are often noisy and incomplete or might not be available at all. With this work, we propose to jointly learn the graph structure and the parameters of graph convolutional networks (GCNs) by approximately solving a bilevel program that learns a discrete probability distribution on the edges of the graph. This allows one to apply GCNs not only in scenarios where the given graph is incomplete or corrupted but also in those where a graph is not available. We conduct a series of experiments that analyze the behavior of the proposed method and demonstrate that it outperforms related methods by a significant margin.'
volume: 97
URL: https://proceedings.mlr.press/v97/franceschi19a.html
PDF: http://proceedings.mlr.press/v97/franceschi19a/franceschi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-franceschi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Luca
family: Franceschi
- given: Mathias
family: Niepert
- given: Massimiliano
family: Pontil
- given: Xiao
family: He
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1972-1982
id: franceschi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1972
lastpage: 1982
published: 2019-05-24 00:00:00 +0000
- title: 'Distributional Multivariate Policy Evaluation and Exploration with the Bellman GAN'
abstract: 'The recently proposed distributional approach to reinforcement learning (DiRL) is centered on learning the distribution of the reward-to-go, often referred to as the value distribution. In this work, we show that the distributional Bellman equation, which drives DiRL methods, is equivalent to a generative adversarial network (GAN) model. In this formulation, DiRL can be seen as learning a deep generative model of the value distribution, driven by the discrepancy between the distribution of the current value, and the distribution of the sum of current reward and next value. We use this insight to propose a GAN-based approach to DiRL, which leverages the strengths of GANs in learning distributions of high dimensional data. In particular, we show that our GAN approach can be used for DiRL with multivariate rewards, an important setting which cannot be tackled with prior methods. The multivariate setting also allows us to unify learning the distribution of values and state transitions, and we exploit this idea to devise a novel exploration method that is driven by the discrepancy in estimating both values and states.'
volume: 97
URL: https://proceedings.mlr.press/v97/freirich19a.html
PDF: http://proceedings.mlr.press/v97/freirich19a/freirich19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-freirich19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Dror
family: Freirich
- given: Tzahi
family: Shimkin
- given: Ron
family: Meir
- given: Aviv
family: Tamar
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1983-1992
id: freirich19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1983
lastpage: 1992
published: 2019-05-24 00:00:00 +0000
- title: 'Approximating Orthogonal Matrices with Effective Givens Factorization'
abstract: 'We analyze effective approximation of unitary matrices. In our formulation, a unitary matrix is represented as a product of rotations in two-dimensional subspaces, so-called Givens rotations. Instead of the quadratic dimension dependence when applying a dense matrix, applying such an approximation scales with the number factors, each of which can be implemented efficiently. Consequently, in settings where an approximation is once computed and then applied many times, such a representation becomes advantageous. Although effective Givens factorization is not possible for generic unitary operators, we show that minimizing a sparsity-inducing objective with a coordinate descent algorithm on the unitary group yields good factorizations for structured matrices. Canonical applications of such a setup are orthogonal basis transforms. We demonstrate numerical results of approximating the graph Fourier transform, which is the matrix obtained when diagonalizing a graph Laplacian.'
volume: 97
URL: https://proceedings.mlr.press/v97/frerix19a.html
PDF: http://proceedings.mlr.press/v97/frerix19a/frerix19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-frerix19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Thomas
family: Frerix
- given: Joan
family: Bruna
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 1993-2001
id: frerix19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 1993
lastpage: 2001
published: 2019-05-24 00:00:00 +0000
- title: 'Fast and Flexible Inference of Joint Distributions from their Marginals'
abstract: 'Across the social sciences and elsewhere, practitioners frequently have to reason about relationships between random variables, despite lacking joint observations of the variables. This is sometimes called an "ecological" inference; given samples from the marginal distributions of the variables, one attempts to infer their joint distribution. The problem is inherently ill-posed, yet only a few models have been proposed for bringing prior information into the problem, often relying on restrictive or unrealistic assumptions and lacking a unified approach. In this paper, we treat the inference problem generally and propose a unified class of models that encompasses some of those previously proposed while including many new ones. Previous work has relied on either relaxation or approximate inference via MCMC, with the latter known to mix prohibitively slowly for this type of problem. Here we instead give a single exact inference algorithm that works for the entire model class via an efficient fixed point iteration called Dykstra’s method. We investigate empirically both the computational cost of our algorithm and the accuracy of the new models on real datasets, showing favorable performance in both cases and illustrating the impact of increased flexibility in modeling enabled by this work.'
volume: 97
URL: https://proceedings.mlr.press/v97/frogner19a.html
PDF: http://proceedings.mlr.press/v97/frogner19a/frogner19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-frogner19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Charlie
family: Frogner
- given: Tomaso
family: Poggio
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2002-2011
id: frogner19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2002
lastpage: 2011
published: 2019-05-24 00:00:00 +0000
- title: 'Analyzing and Improving Representations with the Soft Nearest Neighbor Loss'
abstract: 'We explore and expand the Soft Nearest Neighbor Loss to measure the entanglement of class manifolds in representation space: i.e., how close pairs of points from the same class are relative to pairs of points from different classes. We demonstrate several use cases of the loss. As an analytical tool, it provides insights into the evolution of class similarity structures during learning. Surprisingly, we find that maximizing the entanglement of representations of different classes in the hidden layers is beneficial for discrimination in the final layer, possibly because it encourages representations to identify class-independent similarity structures. Maximizing the soft nearest neighbor loss in the hidden layers leads not only to better-calibrated estimates of uncertainty on outlier data but also marginally improved generalization. Data that is not from the training distribution can be recognized by observing that in the hidden layers, it has fewer than the normal number of neighbors from the predicted class.'
volume: 97
URL: https://proceedings.mlr.press/v97/frosst19a.html
PDF: http://proceedings.mlr.press/v97/frosst19a/frosst19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-frosst19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Nicholas
family: Frosst
- given: Nicolas
family: Papernot
- given: Geoffrey
family: Hinton
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2012-2020
id: frosst19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2012
lastpage: 2020
published: 2019-05-24 00:00:00 +0000
- title: 'Diagnosing Bottlenecks in Deep Q-learning Algorithms'
abstract: 'Q-learning methods are a common class of algorithms used in reinforcement learning (RL). However, their behavior with function approximation, especially with neural networks, is poorly understood theoretically and empirically. In this work, we aim to experimentally investigate potential issues in Q-learning, by means of a "unit testing" framework where we can utilize oracles to disentangle sources of error. Specifically, we investigate questions related to function approximation, sampling error and nonstationarity, and where available, verify if trends found in oracle settings hold true with deep RL methods. We find that large neural network architectures have many benefits with regards to learning stability; offer several practical compensations for overfitting; and develop a novel sampling method based on explicitly compensating for function approximation error that yields fair improvement on high-dimensional continuous control domains.'
volume: 97
URL: https://proceedings.mlr.press/v97/fu19a.html
PDF: http://proceedings.mlr.press/v97/fu19a/fu19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-fu19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Justin
family: Fu
- given: Aviral
family: Kumar
- given: Matthew
family: Soh
- given: Sergey
family: Levine
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2021-2030
id: fu19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2021
lastpage: 2030
published: 2019-05-24 00:00:00 +0000
- title: 'MetricGAN: Generative Adversarial Networks based Black-box Metric Scores Optimization for Speech Enhancement'
abstract: 'Adversarial loss in a conditional generative adversarial network (GAN) is not designed to directly optimize evaluation metrics of a target task, and thus, may not always guide the generator in a GAN to generate data with improved metric scores. To overcome this issue, we propose a novel MetricGAN approach with an aim to optimize the generator with respect to one or multiple evaluation metrics. Moreover, based on MetricGAN, the metric scores of the generated data can also be arbitrarily specified by users. We tested the proposed MetricGAN on a speech enhancement task, which is particularly suitable to verify the proposed approach because there are multiple metrics measuring different aspects of speech signals. Moreover, these metrics are generally complex and could not be fully optimized by Lp or conventional adversarial losses.'
volume: 97
URL: https://proceedings.mlr.press/v97/fu19b.html
PDF: http://proceedings.mlr.press/v97/fu19b/fu19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-fu19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Szu-Wei
family: Fu
- given: Chien-Feng
family: Liao
- given: Yu
family: Tsao
- given: Shou-De
family: Lin
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2031-2041
id: fu19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2031
lastpage: 2041
published: 2019-05-24 00:00:00 +0000
- title: 'Beyond Adaptive Submodularity: Approximation Guarantees of Greedy Policy with Adaptive Submodularity Ratio'
abstract: 'We propose a new concept named adaptive submodularity ratio to study the greedy policy for sequential decision making. While the greedy policy is known to perform well for a wide variety of adaptive stochastic optimization problems in practice, its theoretical properties have been analyzed only for a limited class of problems. We narrow the gap between theory and practice by using adaptive submodularity ratio, which enables us to prove approximation guarantees of the greedy policy for a substantially wider class of problems. Examples of newly analyzed problems include important applications such as adaptive influence maximization and adaptive feature selection. Our adaptive submodularity ratio also provides bounds of adaptivity gaps. Experiments confirm that the greedy policy performs well with the applications being considered compared to standard heuristics.'
volume: 97
URL: https://proceedings.mlr.press/v97/fujii19a.html
PDF: http://proceedings.mlr.press/v97/fujii19a/fujii19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-fujii19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Kaito
family: Fujii
- given: Shinsaku
family: Sakaue
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2042-2051
id: fujii19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2042
lastpage: 2051
published: 2019-05-24 00:00:00 +0000
- title: 'Off-Policy Deep Reinforcement Learning without Exploration'
abstract: 'Many practical applications of reinforcement learning constrain agents to learn from a fixed batch of data which has already been gathered, without offering further possibility for data collection. In this paper, we demonstrate that due to errors introduced by extrapolation, standard off-policy deep reinforcement learning algorithms, such as DQN and DDPG, are incapable of learning with data uncorrelated to the distribution under the current policy, making them ineffective for this fixed batch setting. We introduce a novel class of off-policy algorithms, batch-constrained reinforcement learning, which restricts the action space in order to force the agent towards behaving close to on-policy with respect to a subset of the given data. We present the first continuous control deep reinforcement learning algorithm which can learn effectively from arbitrary, fixed batch data, and empirically demonstrate the quality of its behavior in several tasks.'
volume: 97
URL: https://proceedings.mlr.press/v97/fujimoto19a.html
PDF: http://proceedings.mlr.press/v97/fujimoto19a/fujimoto19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-fujimoto19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Scott
family: Fujimoto
- given: David
family: Meger
- given: Doina
family: Precup
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2052-2062
id: fujimoto19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2052
lastpage: 2062
published: 2019-05-24 00:00:00 +0000
- title: 'Transfer Learning for Related Reinforcement Learning Tasks via Image-to-Image Translation'
abstract: 'Despite the remarkable success of Deep RL in learning control policies from raw pixels, the resulting models do not generalize. We demonstrate that a trained agent fails completely when facing small visual changes, and that fine-tuning—the common transfer learning paradigm—fails to adapt to these changes, to the extent that it is faster to re-train the model from scratch. We show that by separating the visual transfer task from the control policy we achieve substantially better sample efficiency and transfer behavior, allowing an agent trained on the source task to transfer well to the target tasks. The visual mapping from the target to the source domain is performed using unaligned GANs, resulting in a control policy that can be further improved using imitation learning from imperfect demonstrations. We demonstrate the approach on synthetic visual variants of the Breakout game, as well as on transfer between subsequent levels of Road Fighter, a Nintendo car-driving game. A visualization of our approach can be seen in \url{https://youtu.be/4mnkzYyXMn4} and \url{https://youtu.be/KCGTrQi6Ogo}.'
volume: 97
URL: https://proceedings.mlr.press/v97/gamrian19a.html
PDF: http://proceedings.mlr.press/v97/gamrian19a/gamrian19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gamrian19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Shani
family: Gamrian
- given: Yoav
family: Goldberg
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2063-2072
id: gamrian19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2063
lastpage: 2072
published: 2019-05-24 00:00:00 +0000
- title: 'Breaking the Softmax Bottleneck via Learnable Monotonic Pointwise Non-linearities'
abstract: 'The Softmax function on top of a final linear layer is the de facto method to output probability distributions in neural networks. In many applications such as language models or text generation, this model has to produce distributions over large output vocabularies. Recently, this has been shown to have limited representational capacity due to its connection with the rank bottleneck in matrix factorization. However, little is known about the limitations of Linear-Softmax for quantities of practical interest such as cross entropy or mode estimation, a direction that we explore here. As an efficient and effective solution to alleviate this issue, we propose to learn parametric monotonic functions on top of the logits. We theoretically investigate the rank increasing capabilities of such monotonic functions. Empirically, our method improves in two different quality metrics over the traditional Linear-Softmax layer in synthetic and real language model experiments, adding little time or memory overhead, while being comparable to the more computationally expensive mixture of Softmaxes.'
volume: 97
URL: https://proceedings.mlr.press/v97/ganea19a.html
PDF: http://proceedings.mlr.press/v97/ganea19a/ganea19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ganea19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Octavian
family: Ganea
- given: Sylvain
family: Gelly
- given: Gary
family: Becigneul
- given: Aliaksei
family: Severyn
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2073-2082
id: ganea19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2073
lastpage: 2082
published: 2019-05-24 00:00:00 +0000
- title: 'Graph U-Nets'
abstract: 'We consider the problem of representation learning for graph data. Convolutional neural networks can naturally operate on images, but have significant challenges in dealing with graph data. Given images are special cases of graphs with nodes lie on 2D lattices, graph embedding tasks have a natural correspondence with image pixel-wise prediction tasks such as segmentation. While encoder-decoder architectures like U-Nets have been successfully applied on many image pixel-wise prediction tasks, similar methods are lacking for graph data. This is due to the fact that pooling and up-sampling operations are not natural on graph data. To address these challenges, we propose novel graph pooling (gPool) and unpooling (gUnpool) operations in this work. The gPool layer adaptively selects some nodes to form a smaller graph based on their scalar projection values on a trainable projection vector. We further propose the gUnpool layer as the inverse operation of the gPool layer. The gUnpool layer restores the graph into its original structure using the position information of nodes selected in the corresponding gPool layer. Based on our proposed gPool and gUnpool layers, we develop an encoder-decoder model on graph, known as the graph U-Nets. Our experimental results on node classification and graph classification tasks demonstrate that our methods achieve consistently better performance than previous models.'
volume: 97
URL: https://proceedings.mlr.press/v97/gao19a.html
PDF: http://proceedings.mlr.press/v97/gao19a/gao19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gao19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Hongyang
family: Gao
- given: Shuiwang
family: Ji
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2083-2092
id: gao19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2083
lastpage: 2092
published: 2019-05-24 00:00:00 +0000
- title: 'Deep Generative Learning via Variational Gradient Flow'
abstract: 'We propose a framework to learn deep generative models via \textbf{V}ariational \textbf{Gr}adient Fl\textbf{ow} (VGrow) on probability spaces. The evolving distribution that asymptotically converges to the target distribution is governed by a vector field, which is the negative gradient of the first variation of the $f$-divergence between them. We prove that the evolving distribution coincides with the pushforward distribution through the infinitesimal time composition of residual maps that are perturbations of the identity map along the vector field. The vector field depends on the density ratio of the pushforward distribution and the target distribution, which can be consistently learned from a binary classification problem. Connections of our proposed VGrow method with other popular methods, such as VAE, GAN and flow-based methods, have been established in this framework, gaining new insights of deep generative learning. We also evaluated several commonly used divergences, including Kullback-Leibler, Jensen-Shannon, Jeffreys divergences as well as our newly discovered “logD” divergence which serves as the objective function of the logD-trick GAN. Experimental results on benchmark datasets demonstrate that VGrow can generate high-fidelity images in a stable and efficient manner, achieving competitive performance with state-of-the-art GANs.'
volume: 97
URL: https://proceedings.mlr.press/v97/gao19b.html
PDF: http://proceedings.mlr.press/v97/gao19b/gao19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gao19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yuan
family: Gao
- given: Yuling
family: Jiao
- given: Yang
family: Wang
- given: Yao
family: Wang
- given: Can
family: Yang
- given: Shunkang
family: Zhang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2093-2101
id: gao19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2093
lastpage: 2101
published: 2019-05-24 00:00:00 +0000
- title: 'Rate Distortion For Model Compression:From Theory To Practice'
abstract: 'The enormous size of modern deep neural net-works makes it challenging to deploy those models in memory and communication limited scenarios. Thus, compressing a trained model without a significant loss in performance has become an increasingly important task. Tremendous advances has been made recently, where the main technical building blocks are pruning, quantization, and low-rank factorization. In this paper, we propose principled approaches to improve upon the common heuristics used in those building blocks, by studying the fundamental limit for model compression via the rate distortion theory. We prove a lower bound for the rate distortion function for model compression and prove its achievability for linear models. Although this achievable compression scheme is intractable in practice, this analysis motivates a novel objective function for model compression, which can be used to improve classes of model compressor such as pruning or quantization. Theoretically, we prove that the proposed scheme is optimal for compressing one-hidden-layer ReLU neural networks. Empirically,we show that the proposed scheme improves upon the baseline in the compression-accuracy tradeoff.'
volume: 97
URL: https://proceedings.mlr.press/v97/gao19c.html
PDF: http://proceedings.mlr.press/v97/gao19c/gao19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gao19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Weihao
family: Gao
- given: Yu-Han
family: Liu
- given: Chong
family: Wang
- given: Sewoong
family: Oh
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2102-2111
id: gao19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2102
lastpage: 2111
published: 2019-05-24 00:00:00 +0000
- title: 'Demystifying Dropout'
abstract: 'Dropout is a popular technique to train large-scale deep neural networks to alleviate the overfitting problem. To disclose the underlying reasons for its gain, numerous works have tried to explain it from different perspectives. In this paper, unlike existing works, we explore it from a new perspective to provide new insight into this line of research. In detail, we disentangle the forward and backward pass of dropout. Then, we find that these two passes need different levels of noise to improve the generalization performance of deep neural networks. Based on this observation, we propose the augmented dropout which employs different dropping strategies in the forward and backward pass. Experimental results have verified the effectiveness of our proposed method.'
volume: 97
URL: https://proceedings.mlr.press/v97/gao19d.html
PDF: http://proceedings.mlr.press/v97/gao19d/gao19d.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gao19d.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Hongchang
family: Gao
- given: Jian
family: Pei
- given: Heng
family: Huang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2112-2121
id: gao19d
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2112
lastpage: 2121
published: 2019-05-24 00:00:00 +0000
- title: 'Geometric Scattering for Graph Data Analysis'
abstract: 'We explore the generalization of scattering transforms from traditional (e.g., image or audio) signals to graph data, analogous to the generalization of ConvNets in geometric deep learning, and the utility of extracted graph features in graph data analysis. In particular, we focus on the capacity of these features to retain informative variability and relations in the data (e.g., between individual graphs, or in aggregate), while relating our construction to previous theoretical results that establish the stability of similar transforms to families of graph deformations. We demonstrate the application of our geometric scattering features in graph classification of social network data, and in data exploration of biochemistry data.'
volume: 97
URL: https://proceedings.mlr.press/v97/gao19e.html
PDF: http://proceedings.mlr.press/v97/gao19e/gao19e.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gao19e.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Feng
family: Gao
- given: Guy
family: Wolf
- given: Matthew
family: Hirn
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2122-2131
id: gao19e
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2122
lastpage: 2131
published: 2019-05-24 00:00:00 +0000
- title: 'Multi-Frequency Phase Synchronization'
abstract: 'We propose a novel formulation for phase synchronization—the statistical problem of jointly estimating alignment angles from noisy pairwise comparisons—as a nonconvex optimization problem that enforces consistency among the pairwise comparisons in multiple frequency channels. Inspired by harmonic retrieval in signal processing, we develop a simple yet efficient two-stage algorithm that leverages the multi-frequency information. We demonstrate in theory and practice that the proposed algorithm significantly outperforms state-of-the-art phase synchronization algorithms, at a mild computational costs incurred by using the extra frequency channels. We also extend our algorithmic framework to general synchronization problems over compact Lie groups.'
volume: 97
URL: https://proceedings.mlr.press/v97/gao19f.html
PDF: http://proceedings.mlr.press/v97/gao19f/gao19f.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gao19f.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Tingran
family: Gao
- given: Zhizhen
family: Zhao
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2132-2141
id: gao19f
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2132
lastpage: 2141
published: 2019-05-24 00:00:00 +0000
- title: 'Optimal Mini-Batch and Step Sizes for SAGA'
abstract: 'Recently it has been shown that the step sizes of a family of variance reduced gradient methods called the JacSketch methods depend on the expected smoothness constant. In particular, if this expected smoothness constant could be calculated a priori, then one could safely set much larger step sizes which would result in a much faster convergence rate. We fill in this gap, and provide simple closed form expressions for the expected smoothness constant and careful numerical experiments verifying these bounds. Using these bounds, and since the SAGA algorithm is part of this JacSketch family, we suggest a new standard practice for setting the step and mini-batch sizes for SAGA that are competitive with a numerical grid search. Furthermore, we can now show that the total complexity of the SAGA algorithm decreases linearly in the mini-batch size up to a pre-defined value: the optimal mini-batch size. This is a rare result in the stochastic variance reduced literature, only previously shown for the Katyusha algorithm. Finally we conjecture that this is the case for many other stochastic variance reduced methods and that our bounds and analysis of the expected smoothness constant is key to extending these results.'
volume: 97
URL: https://proceedings.mlr.press/v97/gazagnadou19a.html
PDF: http://proceedings.mlr.press/v97/gazagnadou19a/gazagnadou19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gazagnadou19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Nidham
family: Gazagnadou
- given: Robert
family: Gower
- given: Joseph
family: Salmon
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2142-2150
id: gazagnadou19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2142
lastpage: 2150
published: 2019-05-24 00:00:00 +0000
- title: 'SelectiveNet: A Deep Neural Network with an Integrated Reject Option'
abstract: 'We consider the problem of selective prediction (also known as reject option) in deep neural networks, and introduce SelectiveNet, a deep neural architecture with an integrated reject option. Existing rejection mechanisms are based mostly on a threshold over the prediction confidence of a pre-trained network. In contrast, SelectiveNet is trained to optimize both classification (or regression) and rejection simultaneously, end-to-end. The result is a deep neural network that is optimized over the covered domain. In our experiments, we show a consistently improved risk-coverage trade-off over several well-known classification and regression datasets, thus reaching new state-of-the-art results for deep selective classification.'
volume: 97
URL: https://proceedings.mlr.press/v97/geifman19a.html
PDF: http://proceedings.mlr.press/v97/geifman19a/geifman19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-geifman19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yonatan
family: Geifman
- given: Ran
family: El-Yaniv
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2151-2159
id: geifman19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2151
lastpage: 2159
published: 2019-05-24 00:00:00 +0000
- title: 'A Theory of Regularized Markov Decision Processes'
abstract: 'Many recent successful (deep) reinforcement learning algorithms make use of regularization, generally based on entropy or Kullback-Leibler divergence. We propose a general theory of regularized Markov Decision Processes that generalizes these approaches in two directions: we consider a larger class of regularizers, and we consider the general modified policy iteration approach, encompassing both policy iteration and value iteration. The core building blocks of this theory are a notion of regularized Bellman operator and the Legendre-Fenchel transform, a classical tool of convex optimization. This approach allows for error propagation analyses of general algorithmic schemes of which (possibly variants of) classical algorithms such as Trust Region Policy Optimization, Soft Q-learning, Stochastic Actor Critic or Dynamic Policy Programming are special cases. This also draws connections to proximal convex optimization, especially to Mirror Descent.'
volume: 97
URL: https://proceedings.mlr.press/v97/geist19a.html
PDF: http://proceedings.mlr.press/v97/geist19a/geist19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-geist19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Matthieu
family: Geist
- given: Bruno
family: Scherrer
- given: Olivier
family: Pietquin
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2160-2169
id: geist19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2160
lastpage: 2169
published: 2019-05-24 00:00:00 +0000
- title: 'DeepMDP: Learning Continuous Latent Space Models for Representation Learning'
abstract: 'Many reinforcement learning (RL) tasks provide the agent with high-dimensional observations that can be simplified into low-dimensional continuous states. To formalize this process, we introduce the concept of a \texit{DeepMDP}, a parameterized latent space model that is trained via the minimization of two tractable latent space losses: prediction of rewards and prediction of the distribution over next latent states. We show that the optimization of these objectives guarantees (1) the quality of the embedding function as a representation of the state space and (2) the quality of the DeepMDP as a model of the environment. Our theoretical findings are substantiated by the experimental result that a trained DeepMDP recovers the latent structure underlying high-dimensional observations on a synthetic environment. Finally, we show that learning a DeepMDP as an auxiliary task in the Atari 2600 domain leads to large performance improvements over model-free RL.'
volume: 97
URL: https://proceedings.mlr.press/v97/gelada19a.html
PDF: http://proceedings.mlr.press/v97/gelada19a/gelada19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gelada19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Carles
family: Gelada
- given: Saurabh
family: Kumar
- given: Jacob
family: Buckman
- given: Ofir
family: Nachum
- given: Marc G.
family: Bellemare
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2170-2179
id: gelada19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2170
lastpage: 2179
published: 2019-05-24 00:00:00 +0000
- title: 'Partially Linear Additive Gaussian Graphical Models'
abstract: 'We propose a partially linear additive Gaussian graphical model (PLA-GGM) for the estimation of associations between random variables distorted by observed confounders. Model parameters are estimated using an $L_1$-regularized maximal pseudo-profile likelihood estimator (MaPPLE) for which we prove a $\sqrt{n}$-sparsistency. Importantly, our approach avoids parametric constraints on the effects of confounders on the estimated graphical model structure. Empirically, the PLA-GGM is applied to both synthetic and real-world datasets, demonstrating superior performance compared to competing methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/geng19a.html
PDF: http://proceedings.mlr.press/v97/geng19a/geng19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-geng19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sinong
family: Geng
- given: Minhao
family: Yan
- given: Mladen
family: Kolar
- given: Sanmi
family: Koyejo
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2180-2190
id: geng19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2180
lastpage: 2190
published: 2019-05-24 00:00:00 +0000
- title: 'Learning and Data Selection in Big Datasets'
abstract: 'Finding a dataset of minimal cardinality to characterize the optimal parameters of a model is of paramount importance in machine learning and distributed optimization over a network. This paper investigates the compressibility of large datasets. More specifically, we propose a framework that jointly learns the input-output mapping as well as the most representative samples of the dataset (sufficient dataset). Our analytical results show that the cardinality of the sufficient dataset increases sub-linearly with respect to the original dataset size. Numerical evaluations of real datasets reveal a large compressibility, up to 95%, without a noticeable drop in the learnability performance, measured by the generalization error.'
volume: 97
URL: https://proceedings.mlr.press/v97/ghadikolaei19a.html
PDF: http://proceedings.mlr.press/v97/ghadikolaei19a/ghadikolaei19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ghadikolaei19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Hossein Shokri
family: Ghadikolaei
- given: Hadi
family: Ghauch
- given: Carlo
family: Fischione
- given: Mikael
family: Skoglund
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2191-2200
id: ghadikolaei19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2191
lastpage: 2200
published: 2019-05-24 00:00:00 +0000
- title: 'Improved Parallel Algorithms for Density-Based Network Clustering'
abstract: 'Clustering large-scale networks is a central topic in unsupervised learning with many applications in machine learning and data mining. A classic approach to cluster a network is to identify regions of high edge density, which in the literature is captured by two fundamental problems: the densest subgraph and the $k$-core decomposition problems. We design massively parallel computation (MPC) algorithms for these problems that are considerably faster than prior work. In the case of $k$-core decomposition, our work improves exponentially on the algorithm provided by Esfandiari et al. (ICML’18). Compared to the prior work on densest subgraph presented by Bahmani et al. (VLDB’12, ’14), our result requires quadratically fewer MPC rounds. We complement our analysis with an experimental scalability analysis of our techniques.'
volume: 97
URL: https://proceedings.mlr.press/v97/ghaffari19a.html
PDF: http://proceedings.mlr.press/v97/ghaffari19a/ghaffari19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ghaffari19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Mohsen
family: Ghaffari
- given: Silvio
family: Lattanzi
- given: Slobodan
family: Mitrović
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2201-2210
id: ghaffari19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2201
lastpage: 2210
published: 2019-05-24 00:00:00 +0000
- title: 'Recursive Sketches for Modular Deep Learning'
abstract: 'We present a mechanism to compute a sketch (succinct summary) of how a complex modular deep network processes its inputs. The sketch summarizes essential information about the inputs and outputs of the network and can be used to quickly identify key components and summary statistics of the inputs. Furthermore, the sketch is recursive and can be unrolled to identify sub-components of these components and so forth, capturing a potentially complicated DAG structure. These sketches erase gracefully; even if we erase a fraction of the sketch at random, the remainder still retains the “high-weight” information present in the original sketch. The sketches can also be organized in a repository to implicitly form a “knowledge graph”; it is possible to quickly retrieve sketches in the repository that are related to a sketch of interest; arranged in this fashion, the sketches can also be used to learn emerging concepts by looking for new clusters in sketch space. Finally, in the scenario where we want to learn a ground truth deep network, we show that augmenting input/output pairs with these sketches can theoretically make it easier to do so.'
volume: 97
URL: https://proceedings.mlr.press/v97/ghazi19a.html
PDF: http://proceedings.mlr.press/v97/ghazi19a/ghazi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ghazi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Badih
family: Ghazi
- given: Rina
family: Panigrahy
- given: Joshua
family: Wang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2211-2220
id: ghazi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2211
lastpage: 2220
published: 2019-05-24 00:00:00 +0000
- title: 'An Instability in Variational Inference for Topic Models'
abstract: 'Naive mean field variational methods are the state of-the-art approach to inference in topic modeling. We show that these methods suffer from an instability that can produce misleading conclusions. Namely, for certain regimes of the model parameters, variational inference outputs a non-trivial decomposition into topics. However -for the same parameter values- the data contain no actual information about the true topic decomposition, and the output of the algorithm is uncorrelated with it. In particular, the estimated posterior mean is wrong, and estimated credible regions do not achieve the nominal coverage. We discuss how this instability is remedied by more accurate mean field approximations.'
volume: 97
URL: https://proceedings.mlr.press/v97/ghorbani19a.html
PDF: http://proceedings.mlr.press/v97/ghorbani19a/ghorbani19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ghorbani19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Behrooz
family: Ghorbani
- given: Hamid
family: Javadi
- given: Andrea
family: Montanari
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2221-2231
id: ghorbani19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2221
lastpage: 2231
published: 2019-05-24 00:00:00 +0000
- title: 'An Investigation into Neural Net Optimization via Hessian Eigenvalue Density'
abstract: 'To understand the dynamics of training in deep neural networks, we study the evolution of the Hessian eigenvalue density throughout the optimization process. In non-batch normalized networks, we observe the rapid appearance of large isolated eigenvalues in the spectrum, along with a surprising concentration of the gradient in the corresponding eigenspaces. In a batch normalized network, these two effects are almost absent. We give a theoretical rationale to partially explain these phenomena. As part of this work, we adapt advanced tools from numerical linear algebra that allow scalable and accurate estimation of the entire Hessian spectrum of ImageNet-scale neural networks; this technique may be of independent interest in other applications.'
volume: 97
URL: https://proceedings.mlr.press/v97/ghorbani19b.html
PDF: http://proceedings.mlr.press/v97/ghorbani19b/ghorbani19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ghorbani19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Behrooz
family: Ghorbani
- given: Shankar
family: Krishnan
- given: Ying
family: Xiao
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2232-2241
id: ghorbani19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2232
lastpage: 2241
published: 2019-05-24 00:00:00 +0000
- title: 'Data Shapley: Equitable Valuation of Data for Machine Learning'
abstract: 'As data becomes the fuel driving technological and economic growth, a fundamental challenge is how to quantify the value of data in algorithmic predictions and decisions. For example, in healthcare and consumer markets, it has been suggested that individuals should be compensated for the data that they generate, but it is not clear what is an equitable valuation for individual data. In this work, we develop a principled framework to address data valuation in the context of supervised machine learning. Given a learning algorithm trained on $n$ data points to produce a predictor, we propose data Shapley as a metric to quantify the value of each training datum to the predictor performance. Data Shapley uniquely satisfies several natural properties of equitable data valuation. We develop Monte Carlo and gradient-based methods to efficiently estimate data Shapley values in practical settings where complex learning algorithms, including neural networks, are trained on large datasets. In addition to being equitable, extensive experiments across biomedical, image and synthetic data demonstrate that data Shapley has several other benefits: 1) it is more powerful than the popular leave-one-out or leverage score in providing insight on what data is more valuable for a given learning task; 2) low Shapley value data effectively capture outliers and corruptions; 3) high Shapley value data inform what type of new data to acquire to improve the predictor.'
volume: 97
URL: https://proceedings.mlr.press/v97/ghorbani19c.html
PDF: http://proceedings.mlr.press/v97/ghorbani19c/ghorbani19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ghorbani19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Amirata
family: Ghorbani
- given: James
family: Zou
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2242-2251
id: ghorbani19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2242
lastpage: 2251
published: 2019-05-24 00:00:00 +0000
- title: 'Efficient Dictionary Learning with Gradient Descent'
abstract: 'Randomly initialized first-order optimization algorithms are the method of choice for solving many high-dimensional nonconvex problems in machine learning, yet general theoretical guarantees cannot rule out convergence to critical points of poor objective value. For some highly structured nonconvex problems however, the success of gradient descent can be understood by studying the geometry of the objective. We study one such problem – complete orthogonal dictionary learning, and provide converge guarantees for randomly initialized gradient descent to the neighborhood of a global optimum. The resulting rates scale as low order polynomials in the dimension even though the objective possesses an exponential number of saddle points. This efficient convergence can be viewed as a consequence of negative curvature normal to the stable manifolds associated with saddle points, and we provide evidence that this feature is shared by other nonconvex problems of importance as well.'
volume: 97
URL: https://proceedings.mlr.press/v97/gilboa19a.html
PDF: http://proceedings.mlr.press/v97/gilboa19a/gilboa19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gilboa19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Dar
family: Gilboa
- given: Sam
family: Buchanan
- given: John
family: Wright
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2252-2259
id: gilboa19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2252
lastpage: 2259
published: 2019-05-24 00:00:00 +0000
- title: 'A Tree-Based Method for Fast Repeated Sampling of Determinantal Point Processes'
abstract: 'It is often desirable in recommender systems and other information retrieval applications to provide diverse results, and determinantal point processes (DPPs) have become a popular way to capture the trade-off between the quality of individual results and the diversity of the overall set. However, sampling from a DPP is inherently expensive: if the underlying collection contains N items, then generating each DPP sample requires time linear in N following a one-time preprocessing phase. Additionally, results often need to be personalized to a user, but standard approaches to personalization invalidate the preprocessing, making personalized samples especially expensive. In this work we address both of these shortcomings. First, we propose a new algorithm for generating DPP samples in time logarithmic in N, following a slightly more expensive preprocessing phase. We then extend the algorithm to support arbitrary query-time feature weights, allowing us to generate samples customized to individual users while still retaining logarithmic runtime; experiments show our approach runs over 300 times faster than traditional DPP sampling on collections of 100,000 items for samples of size 10.'
volume: 97
URL: https://proceedings.mlr.press/v97/gillenwater19a.html
PDF: http://proceedings.mlr.press/v97/gillenwater19a/gillenwater19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gillenwater19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jennifer
family: Gillenwater
- given: Alex
family: Kulesza
- given: Zelda
family: Mariet
- given: Sergei
family: Vassilvtiskii
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2260-2268
id: gillenwater19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2260
lastpage: 2268
published: 2019-05-24 00:00:00 +0000
- title: 'Learning to Groove with Inverse Sequence Transformations'
abstract: 'We explore models for translating abstract musical ideas (scores, rhythms) into expressive performances using seq2seq and recurrent variational information bottleneck (VIB) models. Though seq2seq models usually require painstakingly aligned corpora, we show that it is possible to adapt an approach from the Generative Adversarial Network (GAN) literature (e.g. Pix2Pix, Vid2Vid) to sequences, creating large volumes of paired data by performing simple transformations and training generative models to plausibly invert these transformations. Music, and drumming in particular, provides a strong test case for this approach because many common transformations (quantization, removing voices) have clear semantics, and learning to invert them has real-world applications. Focusing on the case of drum set players, we create and release a new dataset for this purpose, containing over 13 hours of recordings by professional drummers aligned with fine-grained timing and dynamics information. We also explore some of the creative potential of these models, demonstrating improvements on state-of-the-art methods for Humanization (instantiating a performance from a musical score).'
volume: 97
URL: https://proceedings.mlr.press/v97/gillick19a.html
PDF: http://proceedings.mlr.press/v97/gillick19a/gillick19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gillick19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jon
family: Gillick
- given: Adam
family: Roberts
- given: Jesse
family: Engel
- given: Douglas
family: Eck
- given: David
family: Bamman
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2269-2279
id: gillick19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2269
lastpage: 2279
published: 2019-05-24 00:00:00 +0000
- title: 'Adversarial Examples Are a Natural Consequence of Test Error in Noise'
abstract: 'Over the last few years, the phenomenon of *adversarial examples* — maliciously constructed inputs that fool trained machine learning models — has captured the attention of the research community, especially when restricted to small modifications of a correctly handled input. Less surprisingly, image classifiers also lack human-level performance on randomly corrupted images, such as images with additive Gaussian noise. In this paper we provide both empirical and theoretical evidence that these are two manifestations of the same underlying phenomenon, and therefore the adversarial robustness and corruption robustness research programs are closely related. This suggests that improving adversarial robustness should go hand in hand with improving performance in the presence of more general and realistic image corruptions. This yields a computationally tractable evaluation metric for defenses to consider: test error in noisy image distributions.'
volume: 97
URL: https://proceedings.mlr.press/v97/gilmer19a.html
PDF: http://proceedings.mlr.press/v97/gilmer19a/gilmer19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gilmer19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Justin
family: Gilmer
- given: Nicolas
family: Ford
- given: Nicholas
family: Carlini
- given: Ekin
family: Cubuk
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2280-2289
id: gilmer19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2280
lastpage: 2289
published: 2019-05-24 00:00:00 +0000
- title: 'Discovering Conditionally Salient Features with Statistical Guarantees'
abstract: 'The goal of feature selection is to identify important features that are relevant to explain a outcome variable. Most of the work in this domain has focused on identifying *globally* relevant features, which are features that are related to the outcome using evidence across the entire dataset. We study a more fine-grained statistical problem: *conditional feature selection*, where a feature may be relevant depending on the values of the other features. For example in genetic association studies, variant $A$ could be associated with the phenotype in the entire dataset, but conditioned on variant $B$ being present it might be independent of the phenotype. In this sense, variant $A$ is globally relevant, but conditioned on $B$ it is no longer locally relevant in that region of the feature space. We present a generalization of the knockoff procedure that performs *conditional feature selection* while controlling a generalization of the false discovery rate (FDR) to the conditional setting. By exploiting the feature/response model-free framework of the knockoffs, the quality of the statistical FDR guarantee is not degraded even when we perform conditional feature selections. We implement this method and present an algorithm that automatically partitions the feature space such that it enhances the differences between selected sets in different regions, and validate the statistical theoretical results with experiments.'
volume: 97
URL: https://proceedings.mlr.press/v97/gimenez19a.html
PDF: http://proceedings.mlr.press/v97/gimenez19a/gimenez19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gimenez19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jaime Roquero
family: Gimenez
- given: James
family: Zou
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2290-2298
id: gimenez19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2290
lastpage: 2298
published: 2019-05-24 00:00:00 +0000
- title: 'Estimating Information Flow in Deep Neural Networks'
abstract: 'We study the estimation of the mutual information I(X;T_$\ell$) between the input X to a deep neural network (DNN) and the output vector T_$\ell$ of its $\ell$-th hidden layer (an “internal representation”). Focusing on feedforward networks with fixed weights and noisy internal representations, we develop a rigorous framework for accurate estimation of I(X;T_$\ell$). By relating I(X;T_$\ell$) to information transmission over additive white Gaussian noise channels, we reveal that compression, i.e. reduction in I(X;T_$\ell$) over the course of training, is driven by progressive geometric clustering of the representations of samples from the same class. Experimental results verify this connection. Finally, we shift focus to purely deterministic DNNs, where I(X;T_$\ell$) is provably vacuous, and show that nevertheless, these models also cluster inputs belonging to the same class. The binning-based approximation of I(X;T_$\ell$) employed in past works to measure compression is identified as a measure of clustering, thus clarifying that these experiments were in fact tracking the same clustering phenomenon. Leveraging the clustering perspective, we provide new evidence that compression and generalization may not be causally related and discuss potential future research ideas.'
volume: 97
URL: https://proceedings.mlr.press/v97/goldfeld19a.html
PDF: http://proceedings.mlr.press/v97/goldfeld19a/goldfeld19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-goldfeld19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ziv
family: Goldfeld
- given: Ewout
family: Van Den Berg
- given: Kristjan
family: Greenewald
- given: Igor
family: Melnyk
- given: Nam
family: Nguyen
- given: Brian
family: Kingsbury
- given: Yury
family: Polyanskiy
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2299-2308
id: goldfeld19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2299
lastpage: 2308
published: 2019-05-24 00:00:00 +0000
- title: 'Amortized Monte Carlo Integration'
abstract: 'Current approaches to amortizing Bayesian inference focus solely on approximating the posterior distribution. Typically, this approximation is, in turn, used to calculate expectations for one or more target functions{—}a computational pipeline which is inefficient when the target function(s) are known upfront. In this paper, we address this inefficiency by introducing AMCI, a method for amortizing Monte Carlo integration directly. AMCI operates similarly to amortized inference but produces three distinct amortized proposals, each tailored to a different component of the overall expectation calculation. At runtime, samples are produced separately from each amortized proposal, before being combined to an overall estimate of the expectation. We show that while existing approaches are fundamentally limited in the level of accuracy they can achieve, AMCI can theoretically produce arbitrarily small errors for any integrable target function using only a single sample from each proposal at runtime. We further show that it is able to empirically outperform the theoretically optimal selfnormalized importance sampler on a number of example problems. Furthermore, AMCI allows not only for amortizing over datasets but also amortizing over target functions.'
volume: 97
URL: https://proceedings.mlr.press/v97/golinski19a.html
PDF: http://proceedings.mlr.press/v97/golinski19a/golinski19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-golinski19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Adam
family: Golinski
- given: Frank
family: Wood
- given: Tom
family: Rainforth
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2309-2318
id: golinski19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2309
lastpage: 2318
published: 2019-05-24 00:00:00 +0000
- title: 'Online Algorithms for Rent-Or-Buy with Expert Advice'
abstract: 'We study the use of predictions by multiple experts (such as machine learning algorithms) to improve the performance of online algorithms. In particular, we consider the classical rent-or-buy problem (also called ski rental), and obtain algorithms that provably improve their performance over the adversarial scenario by using these predictions. We also prove matching lower bounds to show that our algorithms are the best possible, and perform experiments to empirically validate their performance in practice'
volume: 97
URL: https://proceedings.mlr.press/v97/gollapudi19a.html
PDF: http://proceedings.mlr.press/v97/gollapudi19a/gollapudi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gollapudi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sreenivas
family: Gollapudi
- given: Debmalya
family: Panigrahi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2319-2327
id: gollapudi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2319
lastpage: 2327
published: 2019-05-24 00:00:00 +0000
- title: 'The information-theoretic value of unlabeled data in semi-supervised learning'
abstract: 'We quantify the separation between the numbers of labeled examples required to learn in two settings: Settings with and without the knowledge of the distribution of the unlabeled data. More specifically, we prove a separation by $\Theta(\log n)$ multiplicative factor for the class of projections over the Boolean hypercube of dimension $n$. We prove that there is no separation for the class of all functions on domain of any size. Learning with the knowledge of the distribution (a.k.a. fixed-distribution learning) can be viewed as an idealized scenario of semi-supervised learning where the number of unlabeled data points is so great that the unlabeled distribution is known exactly. For this reason, we call the separation the value of unlabeled data.'
volume: 97
URL: https://proceedings.mlr.press/v97/golovnev19a.html
PDF: http://proceedings.mlr.press/v97/golovnev19a/golovnev19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-golovnev19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alexander
family: Golovnev
- given: David
family: Pal
- given: Balazs
family: Szorenyi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2328-2336
id: golovnev19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2328
lastpage: 2336
published: 2019-05-24 00:00:00 +0000
- title: 'Efficient Training of BERT by Progressively Stacking'
abstract: 'Unsupervised pre-training is popularly used in natural language processing. By designing proper unsupervised prediction tasks, a deep neural network can be trained and shown to be effective in many downstream tasks. As the data is usually adequate, the model for pre-training is generally huge and contains millions of parameters. Therefore, the training efficiency becomes a critical issue even when using high-performance hardware. In this paper, we explore an efficient training method for the state-of-the-art bidirectional Transformer (BERT) model. By visualizing the self-attention distribution of different layers at different positions in a well-trained BERT model, we find that in most layers, the self-attention distribution will concentrate locally around its position and the start-of-sentence token. Motivating from this, we propose the stacking algorithm to transfer knowledge from a shallow model to a deep model; then we apply stacking progressively to accelerate BERT training. The experimental results showed that the models trained by our training strategy achieve similar performance to models trained from scratch, but our algorithm is much faster.'
volume: 97
URL: https://proceedings.mlr.press/v97/gong19a.html
PDF: http://proceedings.mlr.press/v97/gong19a/gong19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gong19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Linyuan
family: Gong
- given: Di
family: He
- given: Zhuohan
family: Li
- given: Tao
family: Qin
- given: Liwei
family: Wang
- given: Tieyan
family: Liu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2337-2346
id: gong19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2337
lastpage: 2346
published: 2019-05-24 00:00:00 +0000
- title: 'Quantile Stein Variational Gradient Descent for Batch Bayesian Optimization'
abstract: 'Batch Bayesian optimization has been shown to be an efficient and successful approach for black-box function optimization, especially when the evaluation of cost function is highly expensive but can be efficiently parallelized. In this paper, we introduce a novel variational framework for batch query optimization, based on the argument that the query batch should be selected to have both high diversity and good worst case performance. This motivates us to introduce a variational objective that combines a quantile-based risk measure (for worst case performance) and entropy regularization (for enforcing diversity). We derive a gradient-based particle-based algorithm for solving our quantile-based variational objective, which generalizes Stein variational gradient descent (SVGD). We evaluate our method on a number of real-world applications and show that it consistently outperforms other recent state-of-the-art batch Bayesian optimization methods. Extensive experimental results indicate that our method achieves better or comparable performance, compared to the existing methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/gong19b.html
PDF: http://proceedings.mlr.press/v97/gong19b/gong19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gong19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Chengyue
family: Gong
- given: Jian
family: Peng
- given: Qiang
family: Liu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2347-2356
id: gong19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2347
lastpage: 2356
published: 2019-05-24 00:00:00 +0000
- title: 'Obtaining Fairness using Optimal Transport Theory'
abstract: 'In the fair classification setup, we recast the links between fairness and predictability in terms of probability metrics. We analyze repair methods based on mapping conditional distributions to the Wasserstein barycenter. We propose a Random Repair which yields a tradeoff between minimal information loss and a certain amount of fairness.'
volume: 97
URL: https://proceedings.mlr.press/v97/gordaliza19a.html
PDF: http://proceedings.mlr.press/v97/gordaliza19a/gordaliza19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gordaliza19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Paula
family: Gordaliza
- given: Eustasio Del
family: Barrio
- given: Gamboa
family: Fabrice
- given: Jean-Michel
family: Loubes
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2357-2365
id: gordaliza19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2357
lastpage: 2365
published: 2019-05-24 00:00:00 +0000
- title: 'Combining parametric and nonparametric models for off-policy evaluation'
abstract: 'We consider a model-based approach to perform batch off-policy evaluation in reinforcement learning. Our method takes a mixture-of-experts approach to combine parametric and non-parametric models of the environment such that the final value estimate has the least expected error. We do so by first estimating the local accuracy of each model and then using a planner to select which model to use at every time step as to minimize the return error estimate along entire trajectories. Across a variety of domains, our mixture-based approach outperforms the individual models alone as well as state-of-the-art importance sampling-based estimators.'
volume: 97
URL: https://proceedings.mlr.press/v97/gottesman19a.html
PDF: http://proceedings.mlr.press/v97/gottesman19a/gottesman19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gottesman19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Omer
family: Gottesman
- given: Yao
family: Liu
- given: Scott
family: Sussex
- given: Emma
family: Brunskill
- given: Finale
family: Doshi-Velez
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2366-2375
id: gottesman19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2366
lastpage: 2375
published: 2019-05-24 00:00:00 +0000
- title: 'Counterfactual Visual Explanations'
abstract: 'In this work, we develop a technique to produce counterfactual visual explanations. Given a ‘query’ image $I$ for which a vision system predicts class $c$, a counterfactual visual explanation identifies how $I$ could change such that the system would output a different specified class $c’$. To do this, we select a ‘distractor’ image $I’$ that the system predicts as class $c’$ and identify spatial regions in $I$ and $I’$ such that replacing the identified region in $I$ with the identified region in $I’$ would push the system towards classifying $I$ as $c’$. We apply our approach to multiple image classification datasets generating qualitative results showcasing the interpretability and discriminativeness of our counterfactual explanations. To explore the effectiveness of our explanations in teaching humans, we present machine teaching experiments for the task of fine-grained bird classification. We find that users trained to distinguish bird species fare better when given access to counterfactual explanations in addition to training examples.'
volume: 97
URL: https://proceedings.mlr.press/v97/goyal19a.html
PDF: http://proceedings.mlr.press/v97/goyal19a/goyal19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-goyal19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yash
family: Goyal
- given: Ziyan
family: Wu
- given: Jan
family: Ernst
- given: Dhruv
family: Batra
- given: Devi
family: Parikh
- given: Stefan
family: Lee
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2376-2384
id: goyal19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2376
lastpage: 2384
published: 2019-05-24 00:00:00 +0000
- title: 'Adaptive Sensor Placement for Continuous Spaces'
abstract: 'We consider the problem of adaptively placing sensors along an interval to detect stochastically-generated events. We present a new formulation of the problem as a continuum-armed bandit problem with feedback in the form of partial observations of realisations of an inhomogeneous Poisson process. We design a solution method by combining Thompson sampling with nonparametric inference via increasingly granular Bayesian histograms and derive an $\tilde{O}(T^{2/3})$ bound on the Bayesian regret in $T$ rounds. This is coupled with the design of an efficent optimisation approach to select actions in polynomial time. In simulations we demonstrate our approach to have substantially lower and less variable regret than competitor algorithms.'
volume: 97
URL: https://proceedings.mlr.press/v97/grant19a.html
PDF: http://proceedings.mlr.press/v97/grant19a/grant19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-grant19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: James
family: Grant
- given: Alexis
family: Boukouvalas
- given: Ryan-Rhys
family: Griffiths
- given: David
family: Leslie
- given: Sattar
family: Vakili
- given: Enrique Munoz
family: De Cote
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2385-2393
id: grant19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2385
lastpage: 2393
published: 2019-05-24 00:00:00 +0000
- title: 'A Statistical Investigation of Long Memory in Language and Music'
abstract: 'Representation and learning of long-range dependencies is a central challenge confronted in modern applications of machine learning to sequence data. Yet despite the prominence of this issue, the basic problem of measuring long-range dependence, either in a given data source or as represented in a trained deep model, remains largely limited to heuristic tools. We contribute a statistical framework for investigating long-range dependence in current applications of deep sequence modeling, drawing on the well-developed theory of long memory stochastic processes. This framework yields testable implications concerning the relationship between long memory in real-world data and its learned representation in a deep learning architecture, which are explored through a semiparametric framework adapted to the high-dimensional setting.'
volume: 97
URL: https://proceedings.mlr.press/v97/greaves-tunnell19a.html
PDF: http://proceedings.mlr.press/v97/greaves-tunnell19a/greaves-tunnell19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-greaves-tunnell19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alexander
family: Greaves-Tunnell
- given: Zaid
family: Harchaoui
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2394-2403
id: greaves-tunnell19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2394
lastpage: 2403
published: 2019-05-24 00:00:00 +0000
- title: 'Automatic Posterior Transformation for Likelihood-Free Inference'
abstract: 'How can one perform Bayesian inference on stochastic simulators with intractable likelihoods? A recent approach is to learn the posterior from adaptively proposed simulations using neural network-based conditional density estimators. However, existing methods are limited to a narrow range of proposal distributions or require importance weighting that can limit performance in practice. Here we present automatic posterior transformation (APT), a new sequential neural posterior estimation method for simulation-based inference. APT can modify the posterior estimate using arbitrary, dynamically updated proposals, and is compatible with powerful flow-based density estimators. It is more flexible, scalable and efficient than previous simulation-based inference techniques. APT can operate directly on high-dimensional time series and image data, opening up new applications for likelihood-free inference.'
volume: 97
URL: https://proceedings.mlr.press/v97/greenberg19a.html
PDF: http://proceedings.mlr.press/v97/greenberg19a/greenberg19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-greenberg19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: David
family: Greenberg
- given: Marcel
family: Nonnenmacher
- given: Jakob
family: Macke
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2404-2414
id: greenberg19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2404
lastpage: 2414
published: 2019-05-24 00:00:00 +0000
- title: 'Learning to Optimize Multigrid PDE Solvers'
abstract: 'Constructing fast numerical solvers for partial differential equations (PDEs) is crucial for many scientific disciplines. A leading technique for solving large-scale PDEs is using multigrid methods. At the core of a multigrid solver is the prolongation matrix, which relates between different scales of the problem. This matrix is strongly problem-dependent, and its optimal construction is critical to the efficiency of the solver. In practice, however, devising multigrid algorithms for new problems often poses formidable challenges. In this paper we propose a framework for learning multigrid solvers. Our method learns a (single) mapping from discretized PDEs to prolongation operators for a broad class of 2D diffusion problems. We train a neural network once for the entire class of PDEs, using an efficient and unsupervised loss function. Our tests demonstrate improved convergence rates compared to the widely used Black-Box multigrid scheme, suggesting that our method successfully learned rules for constructing prolongation matrices.'
volume: 97
URL: https://proceedings.mlr.press/v97/greenfeld19a.html
PDF: http://proceedings.mlr.press/v97/greenfeld19a/greenfeld19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-greenfeld19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Daniel
family: Greenfeld
- given: Meirav
family: Galun
- given: Ronen
family: Basri
- given: Irad
family: Yavneh
- given: Ron
family: Kimmel
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2415-2423
id: greenfeld19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2415
lastpage: 2423
published: 2019-05-24 00:00:00 +0000
- title: 'Multi-Object Representation Learning with Iterative Variational Inference'
abstract: 'Human perception is structured around objects which form the basis for our higher-level cognition and impressive systematic generalization abilities. Yet most work on representation learning focuses on feature learning without even considering multiple objects, or treats segmentation as an (often supervised) preprocessing step. Instead, we argue for the importance of learning to segment and represent objects jointly. We demonstrate that, starting from the simple assumption that a scene is composed of multiple entities, it is possible to learn to segment images into interpretable objects with disentangled representations. Our method learns – without supervision – to inpaint occluded parts, and extrapolates to scenes with more objects and to unseen objects with novel feature combinations. We also show that, due to the use of iterative variational inference, our system is able to learn multi-modal posteriors for ambiguous inputs and extends naturally to sequences.'
volume: 97
URL: https://proceedings.mlr.press/v97/greff19a.html
PDF: http://proceedings.mlr.press/v97/greff19a/greff19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-greff19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Klaus
family: Greff
- given: Raphaël Lopez
family: Kaufman
- given: Rishabh
family: Kabra
- given: Nick
family: Watters
- given: Christopher
family: Burgess
- given: Daniel
family: Zoran
- given: Loic
family: Matthey
- given: Matthew
family: Botvinick
- given: Alexander
family: Lerchner
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2424-2433
id: greff19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2424
lastpage: 2433
published: 2019-05-24 00:00:00 +0000
- title: 'Graphite: Iterative Generative Modeling of Graphs'
abstract: 'Graphs are a fundamental abstraction for modeling relational data. However, graphs are discrete and combinatorial in nature, and learning representations suitable for machine learning tasks poses statistical and computational challenges. In this work, we propose Graphite, an algorithmic framework for unsupervised learning of representations over nodes in large graphs using deep latent variable generative models. Our model parameterizes variational autoencoders (VAE) with graph neural networks, and uses a novel iterative graph refinement strategy inspired by low-rank approximations for decoding. On a wide variety of synthetic and benchmark datasets, Graphite outperforms competing approaches for the tasks of density estimation, link prediction, and node classification. Finally, we derive a theoretical connection between message passing in graph neural networks and mean-field variational inference.'
volume: 97
URL: https://proceedings.mlr.press/v97/grover19a.html
PDF: http://proceedings.mlr.press/v97/grover19a/grover19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-grover19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Aditya
family: Grover
- given: Aaron
family: Zweig
- given: Stefano
family: Ermon
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2434-2444
id: grover19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2434
lastpage: 2444
published: 2019-05-24 00:00:00 +0000
- title: 'Fast Algorithm for Generalized Multinomial Models with Ranking Data'
abstract: 'We develop a framework of generalized multinomial models, which includes both the popular Plackett–Luce model and Bradley–Terry model as special cases. From a theoretical perspective, we prove that the maximum likelihood estimator (MLE) under generalized multinomial models corresponds to the stationary distribution of an inhomogeneous Markov chain uniquely. Based on this property, we propose an iterative algorithm that is easy to implement and interpret, and is guaranteed to converge. Numerical experiments on synthetic data and real data demonstrate the advantages of our Markov chain based algorithm over existing ones. Our algorithm converges to the MLE with fewer iterations and at a faster convergence rate. The new algorithm is readily applicable to problems such as page ranking or sports ranking data.'
volume: 97
URL: https://proceedings.mlr.press/v97/gu19a.html
PDF: http://proceedings.mlr.press/v97/gu19a/gu19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gu19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jiaqi
family: Gu
- given: Guosheng
family: Yin
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2445-2453
id: gu19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2445
lastpage: 2453
published: 2019-05-24 00:00:00 +0000
- title: 'Towards a Deep and Unified Understanding of Deep Neural Models in NLP'
abstract: 'We define a unified information-based measure to provide quantitative explanations on how intermediate layers of deep Natural Language Processing (NLP) models leverage information of input words. Our method advances existing explanation methods by addressing issues in coherency and generality. Explanations generated by using our method are consistent and faithful across different timestamps, layers, and models. We show how our method can be applied to four widely used models in NLP and explain their performances on three real-world benchmark datasets.'
volume: 97
URL: https://proceedings.mlr.press/v97/guan19a.html
PDF: http://proceedings.mlr.press/v97/guan19a/guan19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-guan19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Chaoyu
family: Guan
- given: Xiting
family: Wang
- given: Quanshi
family: Zhang
- given: Runjin
family: Chen
- given: Di
family: He
- given: Xing
family: Xie
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2454-2463
id: guan19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2454
lastpage: 2463
published: 2019-05-24 00:00:00 +0000
- title: 'An Investigation of Model-Free Planning'
abstract: 'The field of reinforcement learning (RL) is facing increasingly challenging domains with combinatorial complexity. For an RL agent to address these challenges, it is essential that it can plan effectively. Prior work has typically utilized an explicit model of the environment, combined with a specific planning algorithm (such as tree search). More recently, a new family of methods have been proposed that learn how to plan, by providing the structure for planning via an inductive bias in the function approximator (such as a tree structured neural network), trained end-to-end by a model-free RL algorithm. In this paper, we go even further, and demonstrate empirically that an entirely model-free approach, without special structure beyond standard neural network components such as convolutional networks and LSTMs, can learn to exhibit many of the characteristics typically associated with a model-based planner. We measure our agent’s effectiveness at planning in terms of its ability to generalize across a combinatorial and irreversible state space, its data efficiency, and its ability to utilize additional thinking time. We find that our agent has many of the characteristics that one might expect to find in a planning algorithm. Furthermore, it exceeds the state-of-the-art in challenging combinatorial domains such as Sokoban and outperforms other model-free approaches that utilize strong inductive biases toward planning.'
volume: 97
URL: https://proceedings.mlr.press/v97/guez19a.html
PDF: http://proceedings.mlr.press/v97/guez19a/guez19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-guez19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Arthur
family: Guez
- given: Mehdi
family: Mirza
- given: Karol
family: Gregor
- given: Rishabh
family: Kabra
- given: Sebastien
family: Racaniere
- given: Theophane
family: Weber
- given: David
family: Raposo
- given: Adam
family: Santoro
- given: Laurent
family: Orseau
- given: Tom
family: Eccles
- given: Greg
family: Wayne
- given: David
family: Silver
- given: Timothy
family: Lillicrap
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2464-2473
id: guez19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2464
lastpage: 2473
published: 2019-05-24 00:00:00 +0000
- title: 'Humor in Word Embeddings: Cockamamie Gobbledegook for Nincompoops'
abstract: 'While humor is often thought to be beyond the reach of Natural Language Processing, we show that several aspects of single-word humor correlate with simple linear directions in Word Embeddings. In particular: (a) the word vectors capture multiple aspects discussed in humor theories from various disciplines; (b) each individual’s sense of humor can be represented by a vector, which can predict differences in people’s senses of humor on new, unrated, words; and (c) upon clustering humor ratings of multiple demographic groups, different humor preferences emerge across the different groups. Humor ratings are taken from the work of Engelthaler and Hills (2017) as well as from an original crowdsourcing study of 120,000 words. Our dataset further includes annotations for the theoretically-motivated humor features we identify.'
volume: 97
URL: https://proceedings.mlr.press/v97/gultchin19a.html
PDF: http://proceedings.mlr.press/v97/gultchin19a/gultchin19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gultchin19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Limor
family: Gultchin
- given: Genevieve
family: Patterson
- given: Nancy
family: Baym
- given: Nathaniel
family: Swinger
- given: Adam
family: Kalai
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2474-2483
id: gultchin19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2474
lastpage: 2483
published: 2019-05-24 00:00:00 +0000
- title: 'Simple Black-box Adversarial Attacks'
abstract: 'We propose an intriguingly simple method for the construction of adversarial images in the black-box setting. In constrast to the white-box scenario, constructing black-box adversarial images has the additional constraint on query budget, and efficient attacks remain an open problem to date. With only the mild assumption of requiring continuous-valued confidence scores, our highly query-efficient algorithm utilizes the following simple iterative principle: we randomly sample a vector from a predefined orthonormal basis and either add or subtract it to the target image. Despite its simplicity, the proposed method can be used for both untargeted and targeted attacks – resulting in previously unprecedented query efficiency in both settings. We demonstrate the efficacy and efficiency of our algorithm on several real world settings including the Google Cloud Vision API. We argue that our proposed algorithm should serve as a strong baseline for future black-box attacks, in particular because it is extremely fast and its implementation requires less than 20 lines of PyTorch code.'
volume: 97
URL: https://proceedings.mlr.press/v97/guo19a.html
PDF: http://proceedings.mlr.press/v97/guo19a/guo19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-guo19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Chuan
family: Guo
- given: Jacob
family: Gardner
- given: Yurong
family: You
- given: Andrew Gordon
family: Wilson
- given: Kilian
family: Weinberger
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2484-2493
id: guo19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2484
lastpage: 2493
published: 2019-05-24 00:00:00 +0000
- title: 'Exploring interpretable LSTM neural networks over multi-variable data'
abstract: 'For recurrent neural networks trained on time series with target and exogenous variables, in addition to accurate prediction, it is also desired to provide interpretable insights into the data. In this paper, we explore the structure of LSTM recurrent neural networks to learn variable-wise hidden states, with the aim to capture different dynamics in multi-variable time series and distinguish the contribution of variables to the prediction. With these variable-wise hidden states, a mixture attention mechanism is proposed to model the generative process of the target. Then we develop associated training methods to jointly learn network parameters, variable and temporal importance w.r.t the prediction of the target variable. Extensive experiments on real datasets demonstrate enhanced prediction performance by capturing the dynamics of different variables. Meanwhile, we evaluate the interpretation results both qualitatively and quantitatively. It exhibits the prospect as an end-to-end framework for both forecasting and knowledge extraction over multi-variable data.'
volume: 97
URL: https://proceedings.mlr.press/v97/guo19b.html
PDF: http://proceedings.mlr.press/v97/guo19b/guo19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-guo19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Tian
family: Guo
- given: Tao
family: Lin
- given: Nino
family: Antulov-Fantulin
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2494-2504
id: guo19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2494
lastpage: 2504
published: 2019-05-24 00:00:00 +0000
- title: 'Learning to Exploit Long-term Relational Dependencies in Knowledge Graphs'
abstract: 'We study the problem of knowledge graph (KG) embedding. A widely-established assumption to this problem is that similar entities are likely to have similar relational roles. However, existing related methods derive KG embeddings mainly based on triple-level learning, which lack the capability of capturing long-term relational dependencies of entities. Moreover, triple-level learning is insufficient for the propagation of semantic information among entities, especially for the case of cross-KG embedding. In this paper, we propose recurrent skipping networks (RSNs), which employ a skipping mechanism to bridge the gaps between entities. RSNs integrate recurrent neural networks (RNNs) with residual learning to efficiently capture the long-term relational dependencies within and between KGs. We design an end-to-end framework to support RSNs on different tasks. Our experimental results showed that RSNs outperformed state-of-the-art embedding-based methods for entity alignment and achieved competitive performance for KG completion.'
volume: 97
URL: https://proceedings.mlr.press/v97/guo19c.html
PDF: http://proceedings.mlr.press/v97/guo19c/guo19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-guo19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Lingbing
family: Guo
- given: Zequn
family: Sun
- given: Wei
family: Hu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2505-2514
id: guo19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2505
lastpage: 2514
published: 2019-05-24 00:00:00 +0000
- title: 'Memory-Optimal Direct Convolutions for Maximizing Classification Accuracy in Embedded Applications'
abstract: 'In the age of Internet of Things (IoT), embedded devices ranging from ARM Cortex M0s with hundreds of KB of RAM to Arduinos with 2KB RAM are expected to perform increasingly sophisticated classification tasks, such as voice and gesture recognition, activity tracking, and biometric security. While convolutional neural networks (CNNs), together with spectrogram preprocessing, are a natural solution to many of these classification tasks, storage of the network’s activations often exceeds the hard memory constraints of embedded platforms. This paper presents memory-optimal direct convolutions as a way to push classification accuracy as high as possible given strict hardware memory constraints at the expense of extra compute. We therefore explore the opposite end of the compute-memory trade-off curve from standard approaches that minimize latency. We validate the memory-optimal CNN technique with an Arduino implementation of the 10-class MNIST classification task, fitting the network specification, weights, and activations entirely within 2KB SRAM and achieving a state-of-the-art classification accuracy for small-scale embedded systems of 99.15%.'
volume: 97
URL: https://proceedings.mlr.press/v97/gural19a.html
PDF: http://proceedings.mlr.press/v97/gural19a/gural19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-gural19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Albert
family: Gural
- given: Boris
family: Murmann
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2515-2524
id: gural19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2515
lastpage: 2524
published: 2019-05-24 00:00:00 +0000
- title: 'IMEXnet A Forward Stable Deep Neural Network'
abstract: 'Deep convolutional neural networks have revolutionized many machine learning and computer vision tasks, however, some remaining key challenges limit their wider use. These challenges include improving the network’s robustness to perturbations of the input image and the limited “field of view” of convolution operators. We introduce the IMEXnet that addresses these challenges by adapting semi-implicit methods for partial differential equations. Compared to similar explicit networks, such as residual networks, our network is more stable, which has recently shown to reduce the sensitivity to small changes in the input features and improve generalization. The addition of an implicit step connects all pixels in each channel of the image and therefore addresses the field of view problem while still being comparable to standard convolutions in terms of the number of parameters and computational complexity. We also present a new dataset for semantic segmentation and demonstrate the effectiveness of our architecture using the NYU Depth dataset.'
volume: 97
URL: https://proceedings.mlr.press/v97/haber19a.html
PDF: http://proceedings.mlr.press/v97/haber19a/haber19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-haber19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Eldad
family: Haber
- given: Keegan
family: Lensink
- given: Eran
family: Treister
- given: Lars
family: Ruthotto
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2525-2534
id: haber19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2525
lastpage: 2534
published: 2019-05-24 00:00:00 +0000
- title: 'On The Power of Curriculum Learning in Training Deep Networks'
abstract: 'Training neural networks is traditionally done by providing a sequence of random mini-batches sampled uniformly from the entire training data. In this work, we analyze the effect of curriculum learning, which involves the non-uniform sampling of mini-batches, on the training of deep networks, and specifically CNNs trained for image recognition. To employ curriculum learning, the training algorithm must resolve 2 problems: (i) sort the training examples by difficulty; (ii) compute a series of mini-batches that exhibit an increasing level of difficulty. We address challenge (i) using two methods: transfer learning from some competitive “teacher" network, and bootstrapping. In our empirical evaluation, both methods show similar benefits in terms of increased learning speed and improved final performance on test data. We address challenge (ii) by investigating different pacing functions to guide the sampling. The empirical investigation includes a variety of network architectures, using images from CIFAR-10, CIFAR-100 and subsets of ImageNet. We conclude with a novel theoretical analysis of curriculum learning, where we show how it effectively modifies the optimization landscape. We then define the concept of an ideal curriculum, and show that under mild conditions it does not change the corresponding global minimum of the optimization function.'
volume: 97
URL: https://proceedings.mlr.press/v97/hacohen19a.html
PDF: http://proceedings.mlr.press/v97/hacohen19a/hacohen19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hacohen19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Guy
family: Hacohen
- given: Daphna
family: Weinshall
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2535-2544
id: hacohen19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2535
lastpage: 2544
published: 2019-05-24 00:00:00 +0000
- title: 'Trading Redundancy for Communication: Speeding up Distributed SGD for Non-convex Optimization'
abstract: 'Communication overhead is one of the key challenges that hinders the scalability of distributed optimization algorithms to train large neural networks. In recent years, there has been a great deal of research to alleviate communication cost by compressing the gradient vector or using local updates and periodic model averaging. In this paper, we advocate the use of redundancy towards communication-efficient distributed stochastic algorithms for non-convex optimization. In particular, we, both theoretically and practically, show that by properly infusing redundancy to the training data with model averaging, it is possible to significantly reduce the number of communication rounds. To be more precise, we show that redundancy reduces residual error in local averaging, thereby reaching the same level of accuracy with fewer rounds of communication as compared with previous algorithms. Empirical studies on CIFAR10, CIFAR100 and ImageNet datasets in a distributed environment complement our theoretical results; they show that our algorithms have additional beneficial aspects including tolerance to failures, as well as greater gradient diversity.'
volume: 97
URL: https://proceedings.mlr.press/v97/haddadpour19a.html
PDF: http://proceedings.mlr.press/v97/haddadpour19a/haddadpour19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-haddadpour19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Farzin
family: Haddadpour
- given: Mohammad Mahdi
family: Kamani
- given: Mehrdad
family: Mahdavi
- given: Viveck
family: Cadambe
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2545-2554
id: haddadpour19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2545
lastpage: 2554
published: 2019-05-24 00:00:00 +0000
- title: 'Learning Latent Dynamics for Planning from Pixels'
abstract: 'Planning has been very successful for control tasks with known environment dynamics. To leverage planning in unknown environments, the agent needs to learn the dynamics from interactions with the world. However, learning dynamics models that are accurate enough for planning has been a long-standing challenge, especially in image-based domains. We propose the Deep Planning Network (PlaNet), a purely model-based agent that learns the environment dynamics from images and chooses actions through fast online planning in latent space. To achieve high performance, the dynamics model must accurately predict the rewards ahead for multiple time steps. We approach this using a latent dynamics model with both deterministic and stochastic transition components. Moreover, we propose a multi-step variational inference objective that we name latent overshooting. Using only pixel observations, our agent solves continuous control tasks with contact dynamics, partial observability, and sparse rewards, which exceed the difficulty of tasks that were previously solved by planning with learned models. PlaNet uses substantially fewer episodes and reaches final performance close to and sometimes higher than strong model-free algorithms.'
volume: 97
URL: https://proceedings.mlr.press/v97/hafner19a.html
PDF: http://proceedings.mlr.press/v97/hafner19a/hafner19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hafner19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Danijar
family: Hafner
- given: Timothy
family: Lillicrap
- given: Ian
family: Fischer
- given: Ruben
family: Villegas
- given: David
family: Ha
- given: Honglak
family: Lee
- given: James
family: Davidson
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2555-2565
id: hafner19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2555
lastpage: 2565
published: 2019-05-24 00:00:00 +0000
- title: 'Neural Separation of Observed and Unobserved Distributions'
abstract: 'Separating mixed distributions is a long standing challenge for machine learning and signal processing. Most current methods either rely on making strong assumptions on the source distributions or rely on having training samples of each source in the mixture. In this work, we introduce a new method—Neural Egg Separation—to tackle the scenario of extracting a signal from an unobserved distribution additively mixed with a signal from an observed distribution. Our method iteratively learns to separate the known distribution from progressively finer estimates of the unknown distribution. In some settings, Neural Egg Separation is initialization sensitive, we therefore introduce Latent Mixture Masking which ensures a good initialization. Extensive experiments on audio and image separation tasks show that our method outperforms current methods that use the same level of supervision, and often achieves similar performance to full supervision.'
volume: 97
URL: https://proceedings.mlr.press/v97/halperin19a.html
PDF: http://proceedings.mlr.press/v97/halperin19a/halperin19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-halperin19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Tavi
family: Halperin
- given: Ariel
family: Ephrat
- given: Yedid
family: Hoshen
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2566-2575
id: halperin19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2566
lastpage: 2575
published: 2019-05-24 00:00:00 +0000
- title: 'Grid-Wise Control for Multi-Agent Reinforcement Learning in Video Game AI'
abstract: 'We consider the problem of multi-agent reinforcement learning (MARL) in video game AI, where the agents are located in a spatial grid-world environment and the number of agents varies both within and across episodes. The challenge is to flexibly control an arbitrary number of agents while achieving effective collaboration. Existing MARL methods usually suffer from the trade-off between these two considerations. To address the issue, we propose a novel architecture that learns a spatial joint representation of all the agents and outputs grid-wise actions. Each agent will be controlled independently by taking the action from the grid it occupies. By viewing the state information as a grid feature map, we employ a convolutional encoder-decoder as the policy network. This architecture naturally promotes agent communication because of the large receptive field provided by the stacked convolutional layers. Moreover, the spatially shared convolutional parameters enable fast parallel exploration that the experiences discovered by one agent can be immediately transferred to others. The proposed method can be conveniently integrated with general reinforcement learning algorithms, e.g., PPO and Q-learning. We demonstrate the effectiveness of the proposed method in extensive challenging multi-agent tasks in StarCraft II.'
volume: 97
URL: https://proceedings.mlr.press/v97/han19a.html
PDF: http://proceedings.mlr.press/v97/han19a/han19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-han19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Lei
family: Han
- given: Peng
family: Sun
- given: Yali
family: Du
- given: Jiechao
family: Xiong
- given: Qing
family: Wang
- given: Xinghai
family: Sun
- given: Han
family: Liu
- given: Tong
family: Zhang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2576-2585
id: han19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2576
lastpage: 2585
published: 2019-05-24 00:00:00 +0000
- title: 'Dimension-Wise Importance Sampling Weight Clipping for Sample-Efficient Reinforcement Learning'
abstract: 'In importance sampling (IS)-based reinforcement learning algorithms such as Proximal Policy Optimization (PPO), IS weights are typically clipped to avoid large variance in learning. However, policy update from clipped statistics induces large bias in tasks with high action dimensions, and bias from clipping makes it difficult to reuse old samples with large IS weights. In this paper, we consider PPO, a representative on-policy algorithm, and propose its improvement by dimension-wise IS weight clipping which separately clips the IS weight of each action dimension to avoid large bias and adaptively controls the IS weight to bound policy update from the current policy. This new technique enables efficient learning for high action-dimensional tasks and reusing of old samples like in off-policy learning to increase the sample efficiency. Numerical results show that the proposed new algorithm outperforms PPO and other RL algorithms in various Open AI Gym tasks.'
volume: 97
URL: https://proceedings.mlr.press/v97/han19b.html
PDF: http://proceedings.mlr.press/v97/han19b/han19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-han19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Seungyul
family: Han
- given: Youngchul
family: Sung
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2586-2595
id: han19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2586
lastpage: 2595
published: 2019-05-24 00:00:00 +0000
- title: 'Complexity of Linear Regions in Deep Networks'
abstract: 'It is well-known that the expressivity of a neural network depends on its architecture, with deeper networks expressing more complex functions. In the case of networks that compute piecewise linear functions, such as those with ReLU activation, the number of distinct linear regions is a natural measure of expressivity. It is possible to construct networks with merely a single region, or for which the number of linear regions grows exponentially with depth; it is not clear where within this range most networks fall in practice, either before or after training. In this paper, we provide a mathematical framework to count the number of linear regions of a piecewise linear network and measure the volume of the boundaries between these regions. In particular, we prove that for networks at initialization, the average number of regions along any one-dimensional subspace grows linearly in the total number of neurons, far below the exponential upper bound. We also find that the average distance to the nearest region boundary at initialization scales like the inverse of the number of neurons. Our theory suggests that, even after training, the number of linear regions is far below exponential, an intuition that matches our empirical observations. We conclude that the practical expressivity of neural networks is likely far below that of the theoretical maximum, and that this gap can be quantified.'
volume: 97
URL: https://proceedings.mlr.press/v97/hanin19a.html
PDF: http://proceedings.mlr.press/v97/hanin19a/hanin19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hanin19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Boris
family: Hanin
- given: David
family: Rolnick
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2596-2604
id: hanin19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2596
lastpage: 2604
published: 2019-05-24 00:00:00 +0000
- title: 'Importance Sampling Policy Evaluation with an Estimated Behavior Policy'
abstract: 'We consider the problem of off-policy evaluation in Markov decision processes. Off-policy evaluation is the task of evaluating the expected return of one policy with data generated by a different, behavior policy. Importance sampling is a technique for off-policy evaluation that re-weights off-policy returns to account for differences in the likelihood of the returns between the two policies. In this paper, we study importance sampling with an estimated behavior policy where the behavior policy estimate comes from the same set of data used to compute the importance sampling estimate. We find that this estimator often lowers the mean squared error of off-policy evaluation compared to importance sampling with the true behavior policy or using a behavior policy that is estimated from a separate data set. Intuitively, estimating the behavior policy in this way corrects for error due to sampling in the action-space. Our empirical results also extend to other popular variants of importance sampling and show that estimating a non-Markovian behavior policy can further lower large-sample mean squared error even when the true behavior policy is Markovian.'
volume: 97
URL: https://proceedings.mlr.press/v97/hanna19a.html
PDF: http://proceedings.mlr.press/v97/hanna19a/hanna19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hanna19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Josiah
family: Hanna
- given: Scott
family: Niekum
- given: Peter
family: Stone
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2605-2613
id: hanna19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2605
lastpage: 2613
published: 2019-05-24 00:00:00 +0000
- title: 'Doubly-Competitive Distribution Estimation'
abstract: 'Distribution estimation is a statistical-learning cornerstone. Its classical *min-max* formulation minimizes the estimation error for the worst distribution, hence under-performs for practical distributions that, like power-law, are often rather simple. Modern research has therefore focused on two frameworks: *structural* estimation that improves learning accuracy by assuming a simple structure of the underlying distribution; and *competitive*, or *instance-optimal*, estimation that achieves the performance of a genie aided estimator up to a small excess error that vanishes as the sample size grows, regardless of the distribution. This paper combines and strengthens the two frameworks. It designs a single estimator whose excess error vanishes both at a universal rate as the sample size grows, as well as when the (unknown) distribution gets simpler. We show that the resulting algorithm significantly improves the performance guarantees for numerous competitive- and structural-estimation results. The algorithm runs in near-linear time and is robust to model misspecification and domain-symbol permutations.'
volume: 97
URL: https://proceedings.mlr.press/v97/hao19a.html
PDF: http://proceedings.mlr.press/v97/hao19a/hao19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hao19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yi
family: Hao
- given: Alon
family: Orlitsky
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2614-2623
id: hao19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2614
lastpage: 2623
published: 2019-05-24 00:00:00 +0000
- title: 'Random Shuffling Beats SGD after Finite Epochs'
abstract: 'A long-standing problem in stochastic optimization is proving that \rsgd, the without-replacement version of \sgd, converges faster than the usual with-replacement \sgd. Building upon \citep{gurbuzbalaban2015random}, we present the *first* (to our knowledge) non-asymptotic results for this problem by proving that after a reasonable number of epochs \rsgd converges faster than \sgd. Specifically, we prove that for strongly convex, second-order smooth functions, the iterates of \rsgd converge to the optimal solution as $\mathcal{O}(\nicefrac{1}{T^2} + \nicefrac{n^3}{T^3})$, where $n$ is the number of components in the objective, and $T$ is number of iterations. This result implies that after $\mathcal{O}(\sqrt{n})$ epochs, \rsgd is *strictly better* than \sgd (which converges as $\mathcal{O}(\nicefrac{1}{T})$). The key step toward showing this better dependence on $T$ is the introduction of $n$ into the bound; and as our analysis shows, in general a dependence on $n$ is unavoidable without further changes. To understand how \rsgd works in practice, we further explore two empirically useful settings: data sparsity and over-parameterization. For sparse data, \rsgd has the rate $\mathcal{O}\left(\frac{1}{T^2}\right)$, again strictly better than \sgd. Under a setting closely related to over-parameterization, \rsgd is shown to converge faster than \sgd after any *arbitrary* number of iterations. Finally, we extend the analysis of \rsgd to smooth non-convex and convex functions.'
volume: 97
URL: https://proceedings.mlr.press/v97/haochen19a.html
PDF: http://proceedings.mlr.press/v97/haochen19a/haochen19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-haochen19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jeff
family: Haochen
- given: Suvrit
family: Sra
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2624-2633
id: haochen19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2624
lastpage: 2633
published: 2019-05-24 00:00:00 +0000
- title: 'Submodular Maximization beyond Non-negativity: Guarantees, Fast Algorithms, and Applications'
abstract: 'It is generally believed that submodular functions–and the more general class of $\gamma$-weakly submodular functions–may only be optimized under the non-negativity assumption $f(S) \geq 0$. In this paper, we show that once the function is expressed as the difference $f = g - c$, where $g$ is monotone, non-negative, and $\gamma$-weakly submodular and $c$ is non-negative modular, then strong approximation guarantees may be obtained. We present an algorithm for maximizing $g - c$ under a $k$-cardinality constraint which produces a random feasible set $S$ such that $\mathbb{E}[g(S) -c(S)] \geq (1 - e^{-\gamma} - \epsilon) g(\opt) - c(\opt)$, whose running time is $O (\frac{n}{\epsilon} \log^2 \frac{1}{\epsilon})$, independent of $k$. We extend these results to the unconstrained setting by describing an algorithm with the same approximation guarantees and faster $O(n \frac{1}{\epsilon} \log\frac{1}{\epsilon})$ runtime. The main techniques underlying our algorithms are two-fold: the use of a surrogate objective which varies the relative importance between $g$ and $c$ throughout the algorithm, and a geometric sweep over possible $\gamma$ values. Our algorithmic guarantees are complemented by a hardness result showing that no polynomial-time algorithm which accesses $g$ through a value oracle can do better. We empirically demonstrate the success of our algorithms by applying them to experimental design on the Boston Housing dataset and directed vertex cover on the Email EU dataset.'
volume: 97
URL: https://proceedings.mlr.press/v97/harshaw19a.html
PDF: http://proceedings.mlr.press/v97/harshaw19a/harshaw19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-harshaw19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Chris
family: Harshaw
- given: Moran
family: Feldman
- given: Justin
family: Ward
- given: Amin
family: Karbasi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2634-2643
id: harshaw19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2634
lastpage: 2643
published: 2019-05-24 00:00:00 +0000
- title: 'Per-Decision Option Discounting'
abstract: 'In order to solve complex problems an agent must be able to reason over a sufficiently long horizon. Temporal abstraction, commonly modeled through options, offers the ability to reason at many timescales, but the horizon length is still determined by the discount factor of the underlying Markov Decision Process. We propose a modification to the options framework that naturally scales the agent’s horizon with option length. We show that the proposed option-step discount controls a bias-variance trade-off, with larger discounts (counter-intuitively) leading to less estimation variance.'
volume: 97
URL: https://proceedings.mlr.press/v97/harutyunyan19a.html
PDF: http://proceedings.mlr.press/v97/harutyunyan19a/harutyunyan19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-harutyunyan19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Anna
family: Harutyunyan
- given: Peter
family: Vrancx
- given: Philippe
family: Hamel
- given: Ann
family: Nowe
- given: Doina
family: Precup
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2644-2652
id: harutyunyan19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2644
lastpage: 2652
published: 2019-05-24 00:00:00 +0000
- title: 'Submodular Observation Selection and Information Gathering for Quadratic Models'
abstract: 'We study the problem of selecting most informative subset of a large observation set to enable accurate estimation of unknown parameters. This problem arises in a variety of settings in machine learning and signal processing including feature selection, phase retrieval, and target localization. Since for quadratic measurement models the moment matrix of the optimal estimator is generally unknown, majority of prior work resorts to approximation techniques such as linearization of the observation model to optimize the alphabetical optimality criteria of an approximate moment matrix. Conversely, by exploiting a connection to the classical Van Trees’ inequality, we derive new alphabetical optimality criteria without distorting the relational structure of the observation model. We further show that under certain conditions on parameters of the problem these optimality criteria are monotone and (weak) submodular set functions. These results enable us to develop an efficient greedy observation selection algorithm uniquely tailored for quadratic models, and provide theoretical bounds on its achievable utility.'
volume: 97
URL: https://proceedings.mlr.press/v97/hashemi19a.html
PDF: http://proceedings.mlr.press/v97/hashemi19a/hashemi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hashemi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Abolfazl
family: Hashemi
- given: Mahsa
family: Ghasemi
- given: Haris
family: Vikalo
- given: Ufuk
family: Topcu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2653-2662
id: hashemi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2653
lastpage: 2662
published: 2019-05-24 00:00:00 +0000
- title: 'Understanding and Controlling Memory in Recurrent Neural Networks'
abstract: 'To be effective in sequential data processing, Recurrent Neural Networks (RNNs) are required to keep track of past events by creating memories. While the relation between memories and the network’s hidden state dynamics was established over the last decade, previous works in this direction were of a predominantly descriptive nature focusing mainly on locating the dynamical objects of interest. In particular, it remained unclear how dynamical observables affect the performance, how they form and whether they can be manipulated. Here, we utilize different training protocols, datasets and architectures to obtain a range of networks solving a delayed classification task with similar performance, alongside substantial differences in their ability to extrapolate for longer delays. We analyze the dynamics of the network’s hidden state, and uncover the reasons for this difference. Each memory is found to be associated with a nearly steady state of the dynamics which we refer to as a ’slow point’. Slow point speeds predict extrapolation performance across all datasets, protocols and architectures tested. Furthermore, by tracking the formation of the slow points we are able to understand the origin of differences between training protocols. Finally, we propose a novel regularization technique that is based on the relation between hidden state speeds and memory longevity. Our technique manipulates these speeds, thereby leading to a dramatic improvement in memory robustness over time, and could pave the way for a new class of regularization methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/haviv19a.html
PDF: http://proceedings.mlr.press/v97/haviv19a/haviv19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-haviv19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Doron
family: Haviv
- given: Alexander
family: Rivkind
- given: Omri
family: Barak
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2663-2671
id: haviv19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2663
lastpage: 2671
published: 2019-05-24 00:00:00 +0000
- title: 'On the Impact of the Activation function on Deep Neural Networks Training'
abstract: 'The weight initialization and the activation function of deep neural networks have a crucial impact on the performance of the training procedure. An inappropriate selection can lead to the loss of information of the input during forward propagation and the exponential vanishing/exploding of gradients during back-propagation. Understanding the theoretical properties of untrained random networks is key to identifying which deep networks may be trained successfully as recently demonstrated by Samuel et al. (2017) who showed that for deep feedforward neural networks only a specific choice of hyperparameters known as the ‘Edge of Chaos’ can lead to good performance. While the work by Samuel et al. (2017) discuss trainability issues, we focus here on training acceleration and overall performance. We give a comprehensive theoretical analysis of the Edge of Chaos and show that we can indeed tune the initialization parameters and the activation function in order to accelerate the training and improve the performance.'
volume: 97
URL: https://proceedings.mlr.press/v97/hayou19a.html
PDF: http://proceedings.mlr.press/v97/hayou19a/hayou19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hayou19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Soufiane
family: Hayou
- given: Arnaud
family: Doucet
- given: Judith
family: Rousseau
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2672-2680
id: hayou19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2672
lastpage: 2680
published: 2019-05-24 00:00:00 +0000
- title: 'Provably Efficient Maximum Entropy Exploration'
abstract: 'Suppose an agent is in a (possibly unknown) Markov Decision Process in the absence of a reward signal, what might we hope that an agent can efficiently learn to do? This work studies a broad class of objectives that are defined solely as functions of the state-visitation frequencies that are induced by how the agent behaves. For example, one natural, intrinsically defined, objective problem is for the agent to learn a policy which induces a distribution over state space that is as uniform as possible, which can be measured in an entropic sense. We provide an efficient algorithm to optimize such such intrinsically defined objectives, when given access to a black box planning oracle (which is robust to function approximation). Furthermore, when restricted to the tabular setting where we have sample based access to the MDP, our proposed algorithm is provably efficient, both in terms of its sample and computational complexities. Key to our algorithmic methodology is utilizing the conditional gradient method (a.k.a. the Frank-Wolfe algorithm) which utilizes an approximate MDP solver.'
volume: 97
URL: https://proceedings.mlr.press/v97/hazan19a.html
PDF: http://proceedings.mlr.press/v97/hazan19a/hazan19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hazan19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Elad
family: Hazan
- given: Sham
family: Kakade
- given: Karan
family: Singh
- given: Abby
family: Van Soest
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2681-2691
id: hazan19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2681
lastpage: 2691
published: 2019-05-24 00:00:00 +0000
- title: 'On the Long-term Impact of Algorithmic Decision Policies: Effort Unfairness and Feature Segregation through Social Learning'
abstract: 'Most existing notions of algorithmic fairness are one-shot: they ensure some form of allocative equality at the time of decision making, but do not account for the adverse impact of the algorithmic decisions today on the long-term welfare and prosperity of certain segments of the population. We take a broader perspective on algorithmic fairness. We propose an effort-based measure of fairness and present a data-driven framework for characterizing the long-term impact of algorithmic policies on reshaping the underlying population. Motivated by the psychological literature on social learning and the economic literature on equality of opportunity, we propose a micro-scale model of how individuals may respond to decision-making algorithms. We employ existing measures of segregation from sociology and economics to quantify the resulting macro- scale population-level change. Importantly, we observe that different models may shift the group- conditional distribution of qualifications in different directions. Our findings raise a number of important questions regarding the formalization of fairness for decision-making models.'
volume: 97
URL: https://proceedings.mlr.press/v97/heidari19a.html
PDF: http://proceedings.mlr.press/v97/heidari19a/heidari19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-heidari19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Hoda
family: Heidari
- given: Vedant
family: Nanda
- given: Krishna
family: Gummadi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2692-2701
id: heidari19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2692
lastpage: 2701
published: 2019-05-24 00:00:00 +0000
- title: 'Graph Resistance and Learning from Pairwise Comparisons'
abstract: 'We consider the problem of learning the qualities of a collection of items by performing noisy comparisons among them. Following the standard paradigm, we assume there is a fixed “comparison graph” and every neighboring pair of items in this graph is compared k times according to the Bradley-Terry-Luce model (where the probability than an item wins a comparison is proportional the item quality). We are interested in how the relative error in quality estimation scales with the comparison graph in the regime where k is large. We show that, asymptotically, the relevant graph-theoretic quantity is the square root of the resistance of the comparison graph. Specifically, we provide an algorithm with relative error decay that scales with the square root of the graph resistance, and provide a lower bound showing that (up to log factors) a better scaling is impossible. The performance guarantee of our algorithm, both in terms of the graph and the skewness of the item quality distribution, significantly outperforms earlier results.'
volume: 97
URL: https://proceedings.mlr.press/v97/hendrickx19a.html
PDF: http://proceedings.mlr.press/v97/hendrickx19a/hendrickx19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hendrickx19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Julien
family: Hendrickx
- given: Alexander
family: Olshevsky
- given: Venkatesh
family: Saligrama
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2702-2711
id: hendrickx19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2702
lastpage: 2711
published: 2019-05-24 00:00:00 +0000
- title: 'Using Pre-Training Can Improve Model Robustness and Uncertainty'
abstract: 'He et al. (2018) have called into question the utility of pre-training by showing that training from scratch can often yield similar performance to pre-training. We show that although pre-training may not improve performance on traditional classification metrics, it improves model robustness and uncertainty estimates. Through extensive experiments on label corruption, class imbalance, adversarial examples, out-of-distribution detection, and confidence calibration, we demonstrate large gains from pre-training and complementary effects with task-specific methods. We show approximately a 10% absolute improvement over the previous state-of-the-art in adversarial robustness. In some cases, using pre-training without task-specific methods also surpasses the state-of-the-art, highlighting the need for pre-training when evaluating future methods on robustness and uncertainty tasks.'
volume: 97
URL: https://proceedings.mlr.press/v97/hendrycks19a.html
PDF: http://proceedings.mlr.press/v97/hendrycks19a/hendrycks19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hendrycks19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Dan
family: Hendrycks
- given: Kimin
family: Lee
- given: Mantas
family: Mazeika
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2712-2721
id: hendrycks19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2712
lastpage: 2721
published: 2019-05-24 00:00:00 +0000
- title: 'Flow++: Improving Flow-Based Generative Models with Variational Dequantization and Architecture Design'
abstract: 'Flow-based generative models are powerful exact likelihood models with efficient sampling and inference. Despite their computational efficiency, flow-based models generally have much worse density modeling performance compared to state-of-the-art autoregressive models. In this paper, we investigate and improve upon three limiting design choices employed by flow-based models in prior work: the use of uniform noise for dequantization, the use of inexpressive affine flows, and the use of purely convolutional conditioning networks in coupling layers. Based on our findings, we propose Flow++, a new flow-based model that is now the state-of-the-art non-autoregressive model for unconditional density estimation on standard image benchmarks. Our work has begun to close the significant performance gap that has so far existed between autoregressive models and flow-based models.'
volume: 97
URL: https://proceedings.mlr.press/v97/ho19a.html
PDF: http://proceedings.mlr.press/v97/ho19a/ho19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ho19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jonathan
family: Ho
- given: Xi
family: Chen
- given: Aravind
family: Srinivas
- given: Yan
family: Duan
- given: Pieter
family: Abbeel
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2722-2730
id: ho19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2722
lastpage: 2730
published: 2019-05-24 00:00:00 +0000
- title: 'Population Based Augmentation: Efficient Learning of Augmentation Policy Schedules'
abstract: 'A key challenge in leveraging data augmentation for neural network training is choosing an effective augmentation policy from a large search space of candidate operations. Properly chosen augmentation policies can lead to significant generalization improvements; however, state-of-the-art approaches such as AutoAugment are computationally infeasible to run for the ordinary user. In this paper, we introduce a new data augmentation algorithm, Population Based Augmentation (PBA), which generates nonstationary augmentation policy schedules instead of a fixed augmentation policy. We show that PBA can match the performance of AutoAugment on CIFAR-10, CIFAR-100, and SVHN, with three orders of magnitude less overall compute. On CIFAR-10 we achieve a mean test error of 1.46%, which is a slight improvement upon the current state-of-the-art. The code for PBA is open source and is available at https://github.com/arcelien/pba.'
volume: 97
URL: https://proceedings.mlr.press/v97/ho19b.html
PDF: http://proceedings.mlr.press/v97/ho19b/ho19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ho19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Daniel
family: Ho
- given: Eric
family: Liang
- given: Xi
family: Chen
- given: Ion
family: Stoica
- given: Pieter
family: Abbeel
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2731-2741
id: ho19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2731
lastpage: 2741
published: 2019-05-24 00:00:00 +0000
- title: 'Collective Model Fusion for Multiple Black-Box Experts'
abstract: 'Model fusion is a fundamental problem in collec-tive machine learning (ML) where independentexperts with heterogeneous learning architecturesare required to combine expertise to improve pre-dictive performance. This is particularly chal-lenging in information-sensitive domains whereexperts do not have access to each other’s internalarchitecture and local data. This paper presentsthe first collective model fusion framework formultiple experts with heterogeneous black-box ar-chitectures. The proposed method will enable thisby addressing the key issues of how black-boxexperts interact to understand the predictive be-haviors of one another; how these understandingscan be represented and shared efficiently amongthemselves; and how the shared understandingscan be combined to generate high-quality consen-sus prediction. The performance of the resultingframework is analyzed theoretically and demon-strated empirically on several datasets.'
volume: 97
URL: https://proceedings.mlr.press/v97/hoang19a.html
PDF: http://proceedings.mlr.press/v97/hoang19a/hoang19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hoang19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Minh
family: Hoang
- given: Nghia
family: Hoang
- given: Bryan Kian Hsiang
family: Low
- given: Carleton
family: Kingsford
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2742-2750
id: hoang19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2742
lastpage: 2750
published: 2019-05-24 00:00:00 +0000
- title: 'Connectivity-Optimized Representation Learning via Persistent Homology'
abstract: 'We study the problem of learning representations with controllable connectivity properties. This is beneficial in situations when the imposed structure can be leveraged upstream. In particular, we control the connectivity of an autoencoder’s latent space via a novel type of loss, operating on information from persistent homology. Under mild conditions, this loss is differentiable and we present a theoretical analysis of the properties induced by the loss. We choose one-class learning as our upstream task and demonstrate that the imposed structure enables informed parameter selection for modeling the in-class distribution via kernel density estimators. Evaluated on computer vision data, these one-class models exhibit competitive performance and, in a low sample size regime, outperform other methods by a large margin. Notably, our results indicate that a single autoencoder, trained on auxiliary (unlabeled) data, yields a mapping into latent space that can be reused across datasets for one-class learning.'
volume: 97
URL: https://proceedings.mlr.press/v97/hofer19a.html
PDF: http://proceedings.mlr.press/v97/hofer19a/hofer19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hofer19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Christoph
family: Hofer
- given: Roland
family: Kwitt
- given: Marc
family: Niethammer
- given: Mandar
family: Dixit
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2751-2760
id: hofer19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2751
lastpage: 2760
published: 2019-05-24 00:00:00 +0000
- title: 'Better generalization with less data using robust gradient descent'
abstract: 'For learning tasks where the data (or losses) may be heavy-tailed, algorithms based on empirical risk minimization may require a substantial number of observations in order to perform well off-sample. In pursuit of stronger performance under weaker assumptions, we propose a technique which uses a cheap and robust iterative estimate of the risk gradient, which can be easily fed into any steepest descent procedure. Finite-sample risk bounds are provided under weak moment assumptions on the loss gradient. The algorithm is simple to implement, and empirical tests using simulations and real-world data illustrate that more efficient and reliable learning is possible without prior knowledge of the loss tails.'
volume: 97
URL: https://proceedings.mlr.press/v97/holland19a.html
PDF: http://proceedings.mlr.press/v97/holland19a/holland19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-holland19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Matthew
family: Holland
- given: Kazushi
family: Ikeda
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2761-2770
id: holland19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2761
lastpage: 2770
published: 2019-05-24 00:00:00 +0000
- title: 'Emerging Convolutions for Generative Normalizing Flows'
abstract: 'Generative flows are attractive because they admit exact likelihood optimization and efficient image synthesis. Recently, Kingma & Dhariwal (2018) demonstrated with Glow that generative flows are capable of generating high quality images. We generalize the 1 {\texttimes} 1 convolutions proposed in Glow to invertible d {\texttimes} d convolutions, which are more flexible since they operate on both channel and spatial axes. We propose two methods to produce invertible convolutions, that have receptive fields identical to standard convolutions: Emerging convolutions are obtained by chaining specific autoregressive convolutions, and periodic convolutions are decoupled in the frequency domain. Our experiments show that the flexibility of d {\texttimes} d convolutions significantly improves the performance of generative flow models on galaxy images, CIFAR10 and ImageNet.'
volume: 97
URL: https://proceedings.mlr.press/v97/hoogeboom19a.html
PDF: http://proceedings.mlr.press/v97/hoogeboom19a/hoogeboom19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hoogeboom19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Emiel
family: Hoogeboom
- given: Rianne
family: Van Den Berg
- given: Max
family: Welling
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2771-2780
id: hoogeboom19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2771
lastpage: 2780
published: 2019-05-24 00:00:00 +0000
- title: 'Nonconvex Variance Reduced Optimization with Arbitrary Sampling'
abstract: 'We provide the first importance sampling variants of variance reduced algorithms for empirical risk minimization with non-convex loss functions. In particular, we analyze non-convex versions of \texttt{SVRG}, \texttt{SAGA} and \texttt{SARAH}. Our methods have the capacity to speed up the training process by an order of magnitude compared to the state of the art on real datasets. Moreover, we also improve upon current mini-batch analysis of these methods by proposing importance sampling for minibatches in this setting. Surprisingly, our approach can in some regimes lead to superlinear speedup with respect to the minibatch size, which is not usually present in stochastic optimization. All the above results follow from a general analysis of the methods which works with *arbitrary sampling*, i.e., fully general randomized strategy for the selection of subsets of examples to be sampled in each iteration. Finally, we also perform a novel importance sampling analysis of \texttt{SARAH} in the convex setting.'
volume: 97
URL: https://proceedings.mlr.press/v97/horvath19a.html
PDF: http://proceedings.mlr.press/v97/horvath19a/horvath19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-horvath19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Samuel
family: Horváth
- given: Peter
family: Richtarik
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2781-2789
id: horvath19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2781
lastpage: 2789
published: 2019-05-24 00:00:00 +0000
- title: 'Parameter-Efficient Transfer Learning for NLP'
abstract: 'Fine-tuning large pretrained models is an effective transfer mechanism in NLP. However, in the presence of many downstream tasks, fine-tuning is parameter inefficient: an entire new model is required for every task. As an alternative, we propose transfer with adapter modules. Adapter modules yield a compact and extensible model; they add only a few trainable parameters per task, and new tasks can be added without revisiting previous ones. The parameters of the original network remain fixed, yielding a high degree of parameter sharing. To demonstrate adapter’s effectiveness, we transfer the recently proposed BERT Transformer model to $26$ diverse text classification tasks, including the GLUE benchmark. Adapters attain near state-of-the-art performance, whilst adding only a few parameters per task. On GLUE, we attain within $0.8%$ of the performance of full fine-tuning, adding only $3.6%$ parameters per task. By contrast, fine-tuning trains $100%$ of the parameters per task.'
volume: 97
URL: https://proceedings.mlr.press/v97/houlsby19a.html
PDF: http://proceedings.mlr.press/v97/houlsby19a/houlsby19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-houlsby19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Neil
family: Houlsby
- given: Andrei
family: Giurgiu
- given: Stanislaw
family: Jastrzebski
- given: Bruna
family: Morrone
- given: Quentin
family: De Laroussilhe
- given: Andrea
family: Gesmundo
- given: Mona
family: Attariyan
- given: Sylvain
family: Gelly
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2790-2799
id: houlsby19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2790
lastpage: 2799
published: 2019-05-24 00:00:00 +0000
- title: 'Stay With Me: Lifetime Maximization Through Heteroscedastic Linear Bandits With Reneging'
abstract: 'Sequential decision making for lifetime maximization is a critical problem in many real-world applications, such as medical treatment and portfolio selection. In these applications, a “reneging” phenomenon, where participants may disengage from future interactions after observing an unsatisfiable outcome, is rather prevalent. To address the above issue, this paper proposes a model of heteroscedastic linear bandits with reneging, which allows each participant to have a distinct “satisfaction level," with any interaction outcome falling short of that level resulting in that participant reneging. Moreover, it allows the variance of the outcome to be context-dependent. Based on this model, we develop a UCB-type policy, namely HR-UCB, and prove that it achieves $\mathcal{O}\big(\sqrt{{T}(\log({T}))^{3}}\big)$ regret. Finally, we validate the performance of HR-UCB via simulations.'
volume: 97
URL: https://proceedings.mlr.press/v97/hsieh19a.html
PDF: http://proceedings.mlr.press/v97/hsieh19a/hsieh19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hsieh19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ping-Chun
family: Hsieh
- given: Xi
family: Liu
- given: Anirban
family: Bhattacharya
- given: P R
family: Kumar
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2800-2809
id: hsieh19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2800
lastpage: 2809
published: 2019-05-24 00:00:00 +0000
- title: 'Finding Mixed Nash Equilibria of Generative Adversarial Networks'
abstract: 'Generative adversarial networks (GANs) are known to achieve the state-of-the-art performance on various generative tasks, but these results come at the expense of a notoriously difficult training phase. Current training strategies typically draw a connection to optimization theory, whose scope is restricted to local convergence due to the presence of non-convexity. In this work, we tackle the training of GANs by rethinking the problem formulation from the mixed Nash Equilibria (NE) perspective. Via a classical lifting trick, we show that essentially all existing GAN objectives can be relaxed into their mixed strategy forms, whose global optima can be solved via sampling, in contrast to the exclusive use of optimization framework in previous work. We further propose a mean-approximation sampling scheme, which allows to systematically exploit methods for bi-affine games to delineate novel, practical training algorithms of GANs. Finally, we provide experimental evidence that our approach yields comparable or superior results to contemporary training algorithms, and outperforms classical methods such as SGD, Adam, and RMSProp.'
volume: 97
URL: https://proceedings.mlr.press/v97/hsieh19b.html
PDF: http://proceedings.mlr.press/v97/hsieh19b/hsieh19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hsieh19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ya-Ping
family: Hsieh
- given: Chen
family: Liu
- given: Volkan
family: Cevher
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2810-2819
id: hsieh19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2810
lastpage: 2819
published: 2019-05-24 00:00:00 +0000
- title: 'Classification from Positive, Unlabeled and Biased Negative Data'
abstract: 'In binary classification, there are situations where negative (N) data are too diverse to be fully labeled and we often resort to positive-unlabeled (PU) learning in these scenarios. However, collecting a non-representative N set that contains only a small portion of all possible N data can often be much easier in practice. This paper studies a novel classification framework which incorporates such biased N (bN) data in PU learning. We provide a method based on empirical risk minimization to address this PUbN classification problem. Our approach can be regarded as a novel example-weighting algorithm, with the weight of each example computed through a preliminary step that draws inspiration from PU learning. We also derive an estimation error bound for the proposed method. Experimental results demonstrate the effectiveness of our algorithm in not only PUbN learning scenarios but also ordinary PU learning scenarios on several benchmark datasets.'
volume: 97
URL: https://proceedings.mlr.press/v97/hsieh19c.html
PDF: http://proceedings.mlr.press/v97/hsieh19c/hsieh19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hsieh19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yu-Guan
family: Hsieh
- given: Gang
family: Niu
- given: Masashi
family: Sugiyama
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2820-2829
id: hsieh19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2820
lastpage: 2829
published: 2019-05-24 00:00:00 +0000
- title: 'Bayesian Deconditional Kernel Mean Embeddings'
abstract: 'Conditional kernel mean embeddings form an attractive nonparametric framework for representing conditional means of functions, describing the observation processes for many complex models. However, the recovery of the original underlying function of interest whose conditional mean was observed is a challenging inference task. We formalize deconditional kernel mean embeddings as a solution to this inverse problem, and show that it can be naturally viewed as a nonparametric Bayes'' rule. Critically, we introduce the notion of task transformed Gaussian processes and establish deconditional kernel means as their posterior predictive mean. This connection provides Bayesian interpretations and uncertainty estimates for deconditional kernel mean embeddings, explains their regularization hyperparameters, and reveals a marginal likelihood for kernel hyperparameter learning. These revelations further enable practical applications such as likelihood-free inference and learning sparse representations for big data.'
volume: 97
URL: https://proceedings.mlr.press/v97/hsu19a.html
PDF: http://proceedings.mlr.press/v97/hsu19a/hsu19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hsu19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Kelvin
family: Hsu
- given: Fabio
family: Ramos
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2830-2838
id: hsu19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2830
lastpage: 2838
published: 2019-05-24 00:00:00 +0000
- title: 'Faster Stochastic Alternating Direction Method of Multipliers for Nonconvex Optimization'
abstract: 'In this paper, we propose a faster stochastic alternating direction method of multipliers (ADMM) for nonconvex optimization by using a new stochastic path-integrated differential estimator (SPIDER), called as SPIDER-ADMM. Moreover, we prove that the SPIDER-ADMM achieves a record-breaking incremental first-order oracle (IFO) complexity for finding an $\epsilon$-approximate solution. As one of major contribution of this paper, we provide a new theoretical analysis framework for nonconvex stochastic ADMM methods with providing the optimal IFO complexity. Based on this new analysis framework, we study the unsolved optimal IFO complexity of the existing non-convex SVRG-ADMM and SAGA-ADMM methods, and prove their the optimal IFO complexity. Thus, the SPIDER-ADMM improves the existing stochastic ADMM methods. Moreover, we extend SPIDER-ADMM to the online setting, and propose a faster online SPIDER-ADMM. Our theoretical analysis also derives the IFO complexity of the online SPIDER-ADMM. Finally, the experimental results on benchmark datasets validate that the proposed algorithms have faster convergence rate than the existing ADMM algorithms for nonconvex optimization.'
volume: 97
URL: https://proceedings.mlr.press/v97/huang19a.html
PDF: http://proceedings.mlr.press/v97/huang19a/huang19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-huang19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Feihu
family: Huang
- given: Songcan
family: Chen
- given: Heng
family: Huang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2839-2848
id: huang19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2839
lastpage: 2848
published: 2019-05-24 00:00:00 +0000
- title: 'Unsupervised Deep Learning by Neighbourhood Discovery'
abstract: 'Deep convolutional neural networks (CNNs) have demonstrated remarkable success in computer vision by supervisedly learning strong visual feature representations. However, training CNNs relies heavily on the availability of exhaustive training data annotations, limiting significantly their deployment and scalability in many application scenarios. In this work, we introduce a generic unsupervised deep learning approach to training deep models without the need for any manual label supervision. Specifically, we progressively discover sample anchored/centred neighbourhoods to reason and learn the underlying class decision boundaries iteratively and accumulatively. Every single neighbourhood is specially formulated so that all the member samples can share the same unseen class labels at high probability for facilitating the extraction of class discriminative feature representations during training. Experiments on image classification show the performance advantages of the proposed method over the state-of-the-art unsupervised learning models on six benchmarks including both coarse-grained and fine-grained object image categorisation.'
volume: 97
URL: https://proceedings.mlr.press/v97/huang19b.html
PDF: http://proceedings.mlr.press/v97/huang19b/huang19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-huang19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jiabo
family: Huang
- given: Qi
family: Dong
- given: Shaogang
family: Gong
- given: Xiatian
family: Zhu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2849-2858
id: huang19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2849
lastpage: 2858
published: 2019-05-24 00:00:00 +0000
- title: 'Detecting Overlapping and Correlated Communities without Pure Nodes: Identifiability and Algorithm'
abstract: 'Many machine learning problems come in the form of networks with relational data between entities, and one of the key unsupervised learning tasks is to detect communities in such a network. We adopt the mixed-membership stochastic blockmodel as the underlying probabilistic model, and give conditions under which the memberships of a subset of nodes can be uniquely identified. Our method starts by constructing a second-order graph moment, which can be shown to converge to a specific product of the true parameters as the size of the network increases. To correctly recover the true membership parameters, we formulate an optimization problem using insights from convex geometry. We show that if the true memberships satisfy a so-called sufficiently scattered condition, then solving the proposed problem correctly identifies the ground truth. We also propose an efficient algorithm for detecting communities, which is significantly faster than prior work and with better convergence properties. Experiments on synthetic and real data justify the validity of the proposed learning framework for network data.'
volume: 97
URL: https://proceedings.mlr.press/v97/huang19c.html
PDF: http://proceedings.mlr.press/v97/huang19c/huang19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-huang19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Kejun
family: Huang
- given: Xiao
family: Fu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2859-2868
id: huang19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2859
lastpage: 2868
published: 2019-05-24 00:00:00 +0000
- title: 'Hierarchical Importance Weighted Autoencoders'
abstract: 'Importance weighted variational inference (Burda et al., 2015) uses multiple i.i.d. samples to have a tighter variational lower bound. We believe a joint proposal has the potential of reducing the number of redundant samples, and introduce a hierarchical structure to induce correlation. The hope is that the proposals would coordinate to make up for the error made by one another to reduce the variance of the importance estimator. Theoretically, we analyze the condition under which convergence of the estimator variance can be connected to convergence of the lower bound. Empirically, we confirm that maximization of the lower bound does implicitly minimize variance. Further analysis shows that this is a result of negative correlation induced by the proposed hierarchical meta sampling scheme, and performance of inference also improves when the number of samples increases.'
volume: 97
URL: https://proceedings.mlr.press/v97/huang19d.html
PDF: http://proceedings.mlr.press/v97/huang19d/huang19d.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-huang19d.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Chin-Wei
family: Huang
- given: Kris
family: Sankaran
- given: Eeshan
family: Dhekane
- given: Alexandre
family: Lacoste
- given: Aaron
family: Courville
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2869-2878
id: huang19d
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2869
lastpage: 2878
published: 2019-05-24 00:00:00 +0000
- title: 'Stable and Fair Classification'
abstract: 'In a recent study, Friedler et al. pointed out that several fair classification algorithms are not stable with respect to variations in the training set – a crucial consideration in several applications. Motivated by their work, we study the problem of designing classification algorithms that are both fair and stable. We propose an extended framework based on fair classification algorithms that are formulated as optimization problems, by introducing a stability-focused regularization term. Theoretically, we prove an additional stability guarantee, that was lacking in fair classification algorithms, and also provide an accuracy guarantee for our extended framework. Our accuracy guarantee can be used to inform the selection of the regularization parameter in our framework. We assess the benefits of our approach empirically by extending several fair classification algorithms that are shown to achieve the best balance between fairness and accuracy over the \textbf{Adult} dataset. Our empirical results show that our extended framework indeed improves the stability at only a slight sacrifice in accuracy.'
volume: 97
URL: https://proceedings.mlr.press/v97/huang19e.html
PDF: http://proceedings.mlr.press/v97/huang19e/huang19e.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-huang19e.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Lingxiao
family: Huang
- given: Nisheeth
family: Vishnoi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2879-2890
id: huang19e
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2879
lastpage: 2890
published: 2019-05-24 00:00:00 +0000
- title: 'Addressing the Loss-Metric Mismatch with Adaptive Loss Alignment'
abstract: 'In most machine learning training paradigms a fixed, often handcrafted, loss function is assumed to be a good proxy for an underlying evaluation metric. In this work we assess this assumption by meta-learning an adaptive loss function to directly optimize the evaluation metric. We propose a sample efficient reinforcement learning approach for adapting the loss dynamically during training. We empirically show how this formulation improves performance by simultaneously optimizing the evaluation metric and smoothing the loss landscape. We verify our method in metric learning and classification scenarios, showing considerable improvements over the state-of-the-art on a diverse set of tasks. Importantly, our method is applicable to a wide range of loss functions and evaluation metrics. Furthermore, the learned policies are transferable across tasks and data, demonstrating the versatility of the method.'
volume: 97
URL: https://proceedings.mlr.press/v97/huang19f.html
PDF: http://proceedings.mlr.press/v97/huang19f/huang19f.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-huang19f.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Chen
family: Huang
- given: Shuangfei
family: Zhai
- given: Walter
family: Talbott
- given: Miguel Bautista
family: Martin
- given: Shih-Yu
family: Sun
- given: Carlos
family: Guestrin
- given: Josh
family: Susskind
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2891-2900
id: huang19f
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2891
lastpage: 2900
published: 2019-05-24 00:00:00 +0000
- title: 'Causal Discovery and Forecasting in Nonstationary Environments with State-Space Models'
abstract: 'In many scientific fields, such as economics and neuroscience, we are often faced with nonstationary time series, and concerned with both finding causal relations and forecasting the values of variables of interest, both of which are particularly challenging in such nonstationary environments. In this paper, we study causal discovery and forecasting for nonstationary time series. By exploiting a particular type of state-space model to represent the processes, we show that nonstationarity helps to identify the causal structure, and that forecasting naturally benefits from learned causal knowledge. Specifically, we allow changes in both causal strengths and noise variances in the nonlinear state-space models, which, interestingly, renders both the causal structure and model parameters identifiable. Given the causal model, we treat forecasting as a problem in Bayesian inference in the causal model, which exploits the time-varying property of the data and adapts to new observations in a principled manner. Experimental results on synthetic and real-world data sets demonstrate the efficacy of the proposed methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/huang19g.html
PDF: http://proceedings.mlr.press/v97/huang19g/huang19g.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-huang19g.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Biwei
family: Huang
- given: Kun
family: Zhang
- given: Mingming
family: Gong
- given: Clark
family: Glymour
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2901-2910
id: huang19g
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2901
lastpage: 2910
published: 2019-05-24 00:00:00 +0000
- title: 'Composing Entropic Policies using Divergence Correction'
abstract: 'Composing skills mastered in one task to solve novel tasks promises dramatic improvements in the data efficiency of reinforcement learning. Here, we analyze two recent works composing behaviors represented in the form of action-value functions and show that they perform poorly in some situations. As part of this analysis, we extend an important generalization of policy improvement to the maximum entropy framework and introduce an algorithm for the practical implementation of successor features in continuous action spaces. Then we propose a novel approach which addresses the failure cases of prior work and, in principle, recovers the optimal policy during transfer. This method works by explicitly learning the (discounted, future) divergence between base policies. We study this approach in the tabular case and on non-trivial continuous control problems with compositional structure and show that it outperforms or matches existing methods across all tasks considered.'
volume: 97
URL: https://proceedings.mlr.press/v97/hunt19a.html
PDF: http://proceedings.mlr.press/v97/hunt19a/hunt19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hunt19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jonathan
family: Hunt
- given: Andre
family: Barreto
- given: Timothy
family: Lillicrap
- given: Nicolas
family: Heess
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2911-2920
id: hunt19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2911
lastpage: 2920
published: 2019-05-24 00:00:00 +0000
- title: 'HexaGAN: Generative Adversarial Nets for Real World Classification'
abstract: 'Most deep learning classification studies assume clean data. However, when dealing with the real world data, we encounter three problems such as 1) missing data, 2) class imbalance, and 3) missing label problems. These problems undermine the performance of a classifier. Various preprocessing techniques have been proposed to mitigate one of these problems, but an algorithm that assumes and resolves all three problems together has not been proposed yet. In this paper, we propose HexaGAN, a generative adversarial network framework that shows promising classification performance for all three problems. We interpret the three problems from a single perspective to solve them jointly. To enable this, the framework consists of six components, which interact with each other. We also devise novel loss functions corresponding to the architecture. The designed loss functions allow us to achieve state-of-the-art imputation performance, with up to a 14% improvement, and to generate high-quality class-conditional data. We evaluate the classification performance (F1-score) of the proposed method with 20% missingness and confirm up to a 5% improvement in comparison with the performance of combinations of state-of-the-art methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/hwang19a.html
PDF: http://proceedings.mlr.press/v97/hwang19a/hwang19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-hwang19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Uiwon
family: Hwang
- given: Dahuin
family: Jung
- given: Sungroh
family: Yoon
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2921-2930
id: hwang19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2921
lastpage: 2930
published: 2019-05-24 00:00:00 +0000
- title: 'Overcoming Mean-Field Approximations in Recurrent Gaussian Process Models'
abstract: 'We identify a new variational inference scheme for dynamical systems whose transition function is modelled by a Gaussian process. Inference in this setting has either employed computationally intensive MCMC methods, or relied on factorisations of the variational posterior. As we demonstrate in our experiments, the factorisation between latent system states and transition function can lead to a miscalibrated posterior and to learning unnecessarily large noise terms. We eliminate this factorisation by explicitly modelling the dependence between state trajectories and the low-rank representation of our Gaussian process posterior. Samples of the latent states can then be tractably generated by conditioning on this representation. The method we obtain gives better predictive performance and more calibrated estimates of the transition function, yet maintains the same time and space complexities as mean-field methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/ialongo19a.html
PDF: http://proceedings.mlr.press/v97/ialongo19a/ialongo19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ialongo19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alessandro Davide
family: Ialongo
- given: Mark
family: Van Der Wilk
- given: James
family: Hensman
- given: Carl Edward
family: Rasmussen
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2931-2940
id: ialongo19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2931
lastpage: 2940
published: 2019-05-24 00:00:00 +0000
- title: 'Learning Structured Decision Problems with Unawareness'
abstract: 'Structured models of decision making often assume an agent is aware of all possible states and actions in advance. This assumption is sometimes untenable. In this paper, we learn Bayesian Decision Networks from both domain exploration and expert assertions in a way which guarantees convergence to optimal behaviour, even when the agent starts unaware of actions or belief variables that are critical to success. Our experiments show that our agent learns optimal behaviour on both small and large decision problems, and that allowing an agent to conserve information upon making new discoveries results in faster convergence.'
volume: 97
URL: https://proceedings.mlr.press/v97/innes19a.html
PDF: http://proceedings.mlr.press/v97/innes19a/innes19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-innes19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Craig
family: Innes
- given: Alex
family: Lascarides
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2941-2950
id: innes19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2941
lastpage: 2950
published: 2019-05-24 00:00:00 +0000
- title: 'Phase transition in PCA with missing data: Reduced signal-to-noise ratio, not sample size!'
abstract: 'How does missing data affect our ability to learn signal structures? It has been shown that learning signal structure in terms of principal components is dependent on the ratio of sample size and dimensionality and that a critical number of observations is needed before learning starts (Biehl and Mietzner, 1993). Here we generalize this analysis to include missing data. Probabilistic principal component analysis is regularly used for estimating signal structures in datasets with missing data. Our analytic result suggest that the effect of missing data is to effectively reduce signal-to-noise ratio rather than - as generally believed - to reduce sample size. The theory predicts a phase transition in the learning curves and this is indeed found both in simulation data and in real datasets.'
volume: 97
URL: https://proceedings.mlr.press/v97/ipsen19a.html
PDF: http://proceedings.mlr.press/v97/ipsen19a/ipsen19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ipsen19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Niels
family: Ipsen
- given: Lars Kai
family: Hansen
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2951-2960
id: ipsen19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2951
lastpage: 2960
published: 2019-05-24 00:00:00 +0000
- title: 'Actor-Attention-Critic for Multi-Agent Reinforcement Learning'
abstract: 'Reinforcement learning in multi-agent scenarios is important for real-world applications but presents challenges beyond those seen in single-agent settings. We present an actor-critic algorithm that trains decentralized policies in multi-agent settings, using centrally computed critics that share an attention mechanism which selects relevant information for each agent at every timestep. This attention mechanism enables more effective and scalable learning in complex multi-agent environments, when compared to recent approaches. Our approach is applicable not only to cooperative settings with shared rewards, but also individualized reward settings, including adversarial settings, as well as settings that do not provide global states, and it makes no assumptions about the action spaces of the agents. As such, it is flexible enough to be applied to most multi-agent learning problems.'
volume: 97
URL: https://proceedings.mlr.press/v97/iqbal19a.html
PDF: http://proceedings.mlr.press/v97/iqbal19a/iqbal19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-iqbal19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Shariq
family: Iqbal
- given: Fei
family: Sha
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2961-2970
id: iqbal19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2961
lastpage: 2970
published: 2019-05-24 00:00:00 +0000
- title: 'Complementary-Label Learning for Arbitrary Losses and Models'
abstract: 'In contrast to the standard classification paradigm where the true class is given to each training pattern, complementary-label learning only uses training patterns each equipped with a complementary label, which only specifies one of the classes that the pattern does not belong to. The goal of this paper is to derive a novel framework of complementary-label learning with an unbiased estimator of the classification risk, for arbitrary losses and models—all existing methods have failed to achieve this goal. Not only is this beneficial for the learning stage, it also makes model/hyper-parameter selection (through cross-validation) possible without the need of any ordinarily labeled validation data, while using any linear/non-linear models or convex/non-convex loss functions. We further improve the risk estimator by a non-negative correction and gradient ascent trick, and demonstrate its superiority through experiments.'
volume: 97
URL: https://proceedings.mlr.press/v97/ishida19a.html
PDF: http://proceedings.mlr.press/v97/ishida19a/ishida19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ishida19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Takashi
family: Ishida
- given: Gang
family: Niu
- given: Aditya
family: Menon
- given: Masashi
family: Sugiyama
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2971-2980
id: ishida19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2971
lastpage: 2980
published: 2019-05-24 00:00:00 +0000
- title: 'Causal Identification under Markov Equivalence: Completeness Results'
abstract: 'Causal effect identification is the task of determining whether a causal distribution is computable from the combination of an observational distribution and substantive knowledge about the domain under investigation. One of the most studied versions of this problem assumes that knowledge is articulated in the form of a fully known causal diagram, which is arguably a strong assumption in many settings. In this paper, we relax this requirement and consider that the knowledge is articulated in the form of an equivalence class of causal diagrams, in particular, a partial ancestral graph (PAG). This is attractive because a PAG can be learned directly from data, and the scientist does not need to commit to a particular, unique diagram. There are different sufficient conditions for identification in PAGs, but none is complete. We derive a complete algorithm for identification given a PAG. This implies that whenever the causal effect is identifiable, the algorithm returns a valid identification expression; alternatively, it will throw a failure condition, which means that the effect is provably not identifiable. We further provide a graphical characterization of non-identifiability of causal effects in PAGs.'
volume: 97
URL: https://proceedings.mlr.press/v97/jaber19a.html
PDF: http://proceedings.mlr.press/v97/jaber19a/jaber19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-jaber19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Amin
family: Jaber
- given: Jiji
family: Zhang
- given: Elias
family: Bareinboim
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2981-2989
id: jaber19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2981
lastpage: 2989
published: 2019-05-24 00:00:00 +0000
- title: 'Learning from a Learner'
abstract: 'In this paper, we propose a novel setting for Inverse Reinforcement Learning (IRL), namely "Learning from a Learner" (LfL). As opposed to standard IRL, it does not consist in learning a reward by observing an optimal agent but from observations of another learning (and thus sub-optimal) agent. To do so, we leverage the fact that the observed agent’s policy is assumed to improve over time. The ultimate goal of this approach is to recover the actual environment’s reward and to allow the observer to outperform the learner. To recover that reward in practice, we propose methods based on the entropy-regularized policy iteration framework. We discuss different approaches to learn solely from trajectories in the state-action space. We demonstrate the genericity of our method by observing agents implementing various reinforcement learning algorithms. Finally, we show that, on both discrete and continuous state/action tasks, the observer’s performance (that optimizes the recovered reward) can surpass those of the observed agent.'
volume: 97
URL: https://proceedings.mlr.press/v97/jacq19a.html
PDF: http://proceedings.mlr.press/v97/jacq19a/jacq19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-jacq19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alexis
family: Jacq
- given: Matthieu
family: Geist
- given: Ana
family: Paiva
- given: Olivier
family: Pietquin
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 2990-2999
id: jacq19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 2990
lastpage: 2999
published: 2019-05-24 00:00:00 +0000
- title: 'Differentially Private Fair Learning'
abstract: 'Motivated by settings in which predictive models may be required to be non-discriminatory with respect to certain attributes (such as race), but even collecting the sensitive attribute may be forbidden or restricted, we initiate the study of fair learning under the constraint of differential privacy. Our first algorithm is a private implementation of the equalized odds post-processing approach of (Hardt et al., 2016). This algorithm is appealingly simple, but must be able to use protected group membership explicitly at test time, which can be viewed as a form of “disparate treatment”. Our second algorithm is a differentially private version of the oracle-efficient in-processing approach of (Agarwal et al., 2018) which is more complex but need not have access to protected group membership at test time. We identify new tradeoffs between fairness, accuracy, and privacy that emerge only when requiring all three properties, and show that these tradeoffs can be milder if group membership may be used at test time. We conclude with a brief experimental evaluation.'
volume: 97
URL: https://proceedings.mlr.press/v97/jagielski19a.html
PDF: http://proceedings.mlr.press/v97/jagielski19a/jagielski19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-jagielski19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Matthew
family: Jagielski
- given: Michael
family: Kearns
- given: Jieming
family: Mao
- given: Alina
family: Oprea
- given: Aaron
family: Roth
- given: Saeed Sharifi
family: -Malvajerdi
- given: Jonathan
family: Ullman
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3000-3008
id: jagielski19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3000
lastpage: 3008
published: 2019-05-24 00:00:00 +0000
- title: 'Sum-of-Squares Polynomial Flow'
abstract: 'Triangular map is a recent construct in probability theory that allows one to transform any source probability density function to any target density function. Based on triangular maps, we propose a general framework for high-dimensional density estimation, by specifying one-dimensional transformations (equivalently conditional densities) and appropriate conditioner networks. This framework (a) reveals the commonalities and differences of existing autoregressive and flow based methods, (b) allows a unified understanding of the limitations and representation power of these recent approaches and, (c) motivates us to uncover a new Sum-of-Squares (SOS) flow that is interpretable, universal, and easy to train. We perform several synthetic experiments on various density geometries to demonstrate the benefits (and short-comings) of such transformations. SOS flows achieve competitive results in simulations and several real-world datasets.'
volume: 97
URL: https://proceedings.mlr.press/v97/jaini19a.html
PDF: http://proceedings.mlr.press/v97/jaini19a/jaini19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-jaini19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Priyank
family: Jaini
- given: Kira A.
family: Selby
- given: Yaoliang
family: Yu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3009-3018
id: jaini19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3009
lastpage: 3018
published: 2019-05-24 00:00:00 +0000
- title: 'DBSCAN++: Towards fast and scalable density clustering'
abstract: 'DBSCAN is a classical density-based clustering procedure with tremendous practical relevance. However, DBSCAN implicitly needs to compute the empirical density for each sample point, leading to a quadratic worst-case time complexity, which is too slow on large datasets. We propose DBSCAN++, a simple modification of DBSCAN which only requires computing the densities for a chosen subset of points. We show empirically that, compared to traditional DBSCAN, DBSCAN++ can provide not only competitive performance but also added robustness in the bandwidth hyperparameter while taking a fraction of the runtime. We also present statistical consistency guarantees showing the trade-off between computational cost and estimation rates. Surprisingly, up to a certain point, we can enjoy the same estimation rates while lowering computational cost, showing that DBSCAN++ is a sub-quadratic algorithm that attains minimax optimal rates for level-set estimation, a quality that may be of independent interest.'
volume: 97
URL: https://proceedings.mlr.press/v97/jang19a.html
PDF: http://proceedings.mlr.press/v97/jang19a/jang19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-jang19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jennifer
family: Jang
- given: Heinrich
family: Jiang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3019-3029
id: jang19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3019
lastpage: 3029
published: 2019-05-24 00:00:00 +0000
- title: 'Learning What and Where to Transfer'
abstract: 'As the application of deep learning has expanded to real-world problems with insufficient volume of training data, transfer learning recently has gained much attention as means of improving the performance in such small-data regime. However, when existing methods are applied between heterogeneous architectures and tasks, it becomes more important to manage their detailed configurations and often requires exhaustive tuning on them for the desired performance. To address the issue, we propose a novel transfer learning approach based on meta-learning that can automatically learn what knowledge to transfer from the source network to where in the target network. Given source and target networks, we propose an efficient training scheme to learn meta-networks that decide (a) which pairs of layers between the source and target networks should be matched for knowledge transfer and (b) which features and how much knowledge from each feature should be transferred. We validate our meta-transfer approach against recent transfer learning methods on various datasets and network architectures, on which our automated scheme significantly outperforms the prior baselines that find “what and where to transfer” in a hand-crafted manner.'
volume: 97
URL: https://proceedings.mlr.press/v97/jang19b.html
PDF: http://proceedings.mlr.press/v97/jang19b/jang19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-jang19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yunhun
family: Jang
- given: Hankook
family: Lee
- given: Sung Ju
family: Hwang
- given: Jinwoo
family: Shin
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3030-3039
id: jang19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3030
lastpage: 3039
published: 2019-05-24 00:00:00 +0000
- title: 'Social Influence as Intrinsic Motivation for Multi-Agent Deep Reinforcement Learning'
abstract: 'We propose a unified mechanism for achieving coordination and communication in Multi-Agent Reinforcement Learning (MARL), through rewarding agents for having causal influence over other agents’ actions. Causal influence is assessed using counterfactual reasoning. At each timestep, an agent simulates alternate actions that it could have taken, and computes their effect on the behavior of other agents. Actions that lead to bigger changes in other agents’ behavior are considered influential and are rewarded. We show that this is equivalent to rewarding agents for having high mutual information between their actions. Empirical results demonstrate that influence leads to enhanced coordination and communication in challenging social dilemma environments, dramatically increasing the learning curves of the deep RL agents, and leading to more meaningful learned communication protocols. The influence rewards for all agents can be computed in a decentralized way by enabling agents to learn a model of other agents using deep neural networks. In contrast, key previous works on emergent communication in the MARL setting were unable to learn diverse policies in a decentralized manner and had to resort to centralized training. Consequently, the influence reward opens up a window of new opportunities for research in this area.'
volume: 97
URL: https://proceedings.mlr.press/v97/jaques19a.html
PDF: http://proceedings.mlr.press/v97/jaques19a/jaques19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-jaques19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Natasha
family: Jaques
- given: Angeliki
family: Lazaridou
- given: Edward
family: Hughes
- given: Caglar
family: Gulcehre
- given: Pedro
family: Ortega
- given: Dj
family: Strouse
- given: Joel Z.
family: Leibo
- given: Nando
family: De Freitas
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3040-3049
id: jaques19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3040
lastpage: 3049
published: 2019-05-24 00:00:00 +0000
- title: 'A Deep Reinforcement Learning Perspective on Internet Congestion Control'
abstract: 'We present and investigate a novel and timely application domain for deep reinforcement learning (RL): Internet congestion control. Congestion control is the core networking task of modulating traffic sources’ data-transmission rates to efficiently utilize network capacity, and is the subject of extensive attention in light of the advent of Internet services such as live video, virtual reality, Internet-of-Things, and more. We show that casting congestion control as RL enables training deep network policies that capture intricate patterns in data traffic and network conditions, and leverage this to outperform the state-of-the-art. We also highlight significant challenges facing real-world adoption of RL-based congestion control, including fairness, safety, and generalization, which are not trivial to address within conventional RL formalism. To facilitate further research and reproducibility of our results, we present a test suite for RL-guided congestion control based on the OpenAI Gym interface.'
volume: 97
URL: https://proceedings.mlr.press/v97/jay19a.html
PDF: http://proceedings.mlr.press/v97/jay19a/jay19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-jay19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Nathan
family: Jay
- given: Noga
family: Rotman
- given: Brighten
family: Godfrey
- given: Michael
family: Schapira
- given: Aviv
family: Tamar
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3050-3059
id: jay19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3050
lastpage: 3059
published: 2019-05-24 00:00:00 +0000
- title: 'Graph Neural Network for Music Score Data and Modeling Expressive Piano Performance'
abstract: 'Music score is often handled as one-dimensional sequential data. Unlike words in a text document, notes in music score can be played simultaneously by the polyphonic nature and each of them has its own duration. In this paper, we represent the unique form of musical score using graph neural network and apply it for rendering expressive piano performance from the music score. Specifically, we design the model using note-level gated graph neural network and measure-level hierarchical attention network with bidirectional long short-term memory with an iterative feedback method. In addition, to model different styles of performance for a given input score, we employ a variational auto-encoder. The result of the listening test shows that our proposed model generated more human-like performances compared to a baseline model and a hierarchical attention network model that handles music score as a word-like sequence.'
volume: 97
URL: https://proceedings.mlr.press/v97/jeong19a.html
PDF: http://proceedings.mlr.press/v97/jeong19a/jeong19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-jeong19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Dasaem
family: Jeong
- given: Taegyun
family: Kwon
- given: Yoojin
family: Kim
- given: Juhan
family: Nam
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3060-3070
id: jeong19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3060
lastpage: 3070
published: 2019-05-24 00:00:00 +0000
- title: 'Ladder Capsule Network'
abstract: 'We propose a new architecture of the capsule network called the ladder capsule network, which has an alternative building block to the dynamic routing algorithm in the capsule network (Sabour et al., 2017). Motivated by the need for using only important capsules during training for robust performance, we first introduce a new layer called the pruning layer, which removes irrelevant capsules. Based on the selected capsules, we construct higher-level capsule outputs. Subsequently, to capture the part-whole spatial relationships, we introduce another new layer called the ladder layer, the outputs of which are regressed lower-level capsule outputs from higher-level capsules. Unlike the capsule network adopting the routing-by-agreement, the ladder capsule network uses backpropagation from a loss function to reconstruct the lower-level capsule outputs from higher-level capsules; thus, the ladder layer implements the reverse directional inference of the agreement/disagreement mechanism of the capsule network. The experiments on MNIST demonstrate that the ladder capsule network learns an equivariant representation and improves the capability to extrapolate or generalize to pose variations.'
volume: 97
URL: https://proceedings.mlr.press/v97/jeong19b.html
PDF: http://proceedings.mlr.press/v97/jeong19b/jeong19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-jeong19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Taewon
family: Jeong
- given: Youngmin
family: Lee
- given: Heeyoung
family: Kim
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3071-3079
id: jeong19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3071
lastpage: 3079
published: 2019-05-24 00:00:00 +0000
- title: 'Training CNNs with Selective Allocation of Channels'
abstract: 'Recent progress in deep convolutional neural networks (CNNs) have enabled a simple paradigm of architecture design: larger models typically achieve better accuracy. Due to this, in modern CNN architectures, it becomes more important to design models that generalize well under certain resource constraints, e.g. the number of parameters. In this paper, we propose a simple way to improve the capacity of any CNN model having large-scale features, without adding more parameters. In particular, we modify a standard convolutional layer to have a new functionality of channel-selectivity, so that the layer is trained to select important channels to re-distribute their parameters. Our experimental results under various CNN architectures and datasets demonstrate that the proposed new convolutional layer allows new optima that generalize better via efficient resource utilization, compared to the baseline.'
volume: 97
URL: https://proceedings.mlr.press/v97/jeong19c.html
PDF: http://proceedings.mlr.press/v97/jeong19c/jeong19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-jeong19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jongheon
family: Jeong
- given: Jinwoo
family: Shin
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3080-3090
id: jeong19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3080
lastpage: 3090
published: 2019-05-24 00:00:00 +0000
- title: 'Learning Discrete and Continuous Factors of Data via Alternating Disentanglement'
abstract: 'We address the problem of unsupervised disentanglement of discrete and continuous explanatory factors of data. We first show a simple procedure for minimizing the total correlation of the continuous latent variables without having to use a discriminator network or perform importance sampling, via cascading the information flow in the beta-VAE framework. Furthermore, we propose a method which avoids offloading the entire burden of jointly modeling the continuous and discrete factors to the variational encoder by employing a separate discrete inference procedure. This leads to an interesting alternating minimization problem which switches between finding the most likely discrete configuration given the continuous factors and updating the variational encoder based on the computed discrete factors. Experiments show that the proposed method clearly disentangles discrete factors and significantly outperforms current disentanglement methods based on the disentanglement score and inference network classification score. The source code is available at https://github.com/snumllab/DisentanglementICML19.'
volume: 97
URL: https://proceedings.mlr.press/v97/jeong19d.html
PDF: http://proceedings.mlr.press/v97/jeong19d/jeong19d.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-jeong19d.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yeonwoo
family: Jeong
- given: Hyun Oh
family: Song
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3091-3099
id: jeong19d
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3091
lastpage: 3099
published: 2019-05-24 00:00:00 +0000
- title: 'Improved Zeroth-Order Variance Reduced Algorithms and Analysis for Nonconvex Optimization'
abstract: 'Two types of zeroth-order stochastic algorithms have recently been designed for nonconvex optimization respectively based on the first-order techniques SVRG and SARAH/SPIDER. This paper addresses several important issues that are still open in these methods. First, all existing SVRG-type zeroth-order algorithms suffer from worse function query complexities than either zeroth-order gradient descent (ZO-GD) or stochastic gradient descent (ZO-SGD). In this paper, we propose a new algorithm ZO-SVRG-Coord-Rand and develop a new analysis for an existing ZO-SVRG-Coord algorithm proposed in Liu et al. 2018b, and show that both ZO-SVRG-Coord-Rand and ZO-SVRG-Coord (under our new analysis) outperform other exiting SVRG-type zeroth-order methods as well as ZO-GD and ZO-SGD. Second, the existing SPIDER-type algorithm SPIDER-SZO (Fang et al., 2018) has superior theoretical performance, but suffers from the generation of a large number of Gaussian random variables as well as a $\sqrt{\epsilon}$-level stepsize in practice. In this paper, we develop a new algorithm ZO-SPIDER-Coord, which is free from Gaussian variable generation and allows a large constant stepsize while maintaining the same convergence rate and query complexity, and we further show that ZO-SPIDER-Coord automatically achieves a linear convergence rate as the iterate enters into a local PL region without restart and algorithmic modification.'
volume: 97
URL: https://proceedings.mlr.press/v97/ji19a.html
PDF: http://proceedings.mlr.press/v97/ji19a/ji19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ji19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Kaiyi
family: Ji
- given: Zhe
family: Wang
- given: Yi
family: Zhou
- given: Yingbin
family: Liang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3100-3109
id: ji19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3100
lastpage: 3109
published: 2019-05-24 00:00:00 +0000
- title: 'Neural Logic Reinforcement Learning'
abstract: 'Deep reinforcement learning (DRL) has achieved significant breakthroughs in various tasks. However, most DRL algorithms suffer a problem of generalising the learned policy, which makes the policy performance largely affected even by minor modifications of the training environment. Except that, the use of deep neural networks makes the learned policies hard to be interpretable. To address these two challenges, we propose a novel algorithm named Neural Logic Reinforcement Learning (NLRL) to represent the policies in reinforcement learning by first-order logic. NLRL is based on policy gradient methods and differentiable inductive logic programming that have demonstrated significant advantages in terms of interpretability and generalisability in supervised tasks. Extensive experiments conducted on cliff-walking and blocks manipulation tasks demonstrate that NLRL can induce interpretable policies achieving near-optimal performance while showing good generalisability to environments of different initial states and problem sizes.'
volume: 97
URL: https://proceedings.mlr.press/v97/jiang19a.html
PDF: http://proceedings.mlr.press/v97/jiang19a/jiang19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-jiang19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Zhengyao
family: Jiang
- given: Shan
family: Luo
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3110-3119
id: jiang19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3110
lastpage: 3119
published: 2019-05-24 00:00:00 +0000
- title: 'Finding Options that Minimize Planning Time'
abstract: 'We formalize the problem of selecting the optimal set of options for planning as that of computing the smallest set of options so that planning converges in less than a given maximum of value-iteration passes. We first show that the problem is $\NP$-hard, even if the task is constrained to be deterministic—the first such complexity result for option discovery. We then present the first polynomial-time boundedly suboptimal approximation algorithm for this setting, and empirically evaluate it against both the optimal options and a representative collection of heuristic approaches in simple grid-based domains.'
volume: 97
URL: https://proceedings.mlr.press/v97/jinnai19a.html
PDF: http://proceedings.mlr.press/v97/jinnai19a/jinnai19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-jinnai19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yuu
family: Jinnai
- given: David
family: Abel
- given: David
family: Hershkowitz
- given: Michael
family: Littman
- given: George
family: Konidaris
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3120-3129
id: jinnai19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3120
lastpage: 3129
published: 2019-05-24 00:00:00 +0000
- title: 'Discovering Options for Exploration by Minimizing Cover Time'
abstract: 'One of the main challenges in reinforcement learning is solving tasks with sparse reward. We show that the difficulty of discovering a distant rewarding state in an MDP is bounded by the expected cover time of a random walk over the graph induced by the MDP’s transition dynamics. We therefore propose to accelerate exploration by constructing options that minimize cover time. We introduce a new option discovery algorithm that diminishes the expected cover time by connecting the most distant states in the state-space graph with options. We show empirically that the proposed algorithm improves learning in several domains with sparse rewards.'
volume: 97
URL: https://proceedings.mlr.press/v97/jinnai19b.html
PDF: http://proceedings.mlr.press/v97/jinnai19b/jinnai19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-jinnai19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yuu
family: Jinnai
- given: Jee Won
family: Park
- given: David
family: Abel
- given: George
family: Konidaris
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3130-3139
id: jinnai19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3130
lastpage: 3139
published: 2019-05-24 00:00:00 +0000
- title: 'Kernel Mean Matching for Content Addressability of GANs'
abstract: 'We propose a novel procedure which adds "content-addressability" to any given unconditional implicit model e.g., a generative adversarial network (GAN). The procedure allows users to control the generative process by specifying a set (arbitrary size) of desired examples based on which similar samples are generated from the model. The proposed approach, based on kernel mean matching, is applicable to any generative models which transform latent vectors to samples, and does not require retraining of the model. Experiments on various high-dimensional image generation problems (CelebA-HQ, LSUN bedroom, bridge, tower) show that our approach is able to generate images which are consistent with the input set, while retaining the image quality of the original model. To our knowledge, this is the first work that attempts to construct, at test time, a content-addressable generative model from a trained marginal model.'
volume: 97
URL: https://proceedings.mlr.press/v97/jitkrittum19a.html
PDF: http://proceedings.mlr.press/v97/jitkrittum19a/jitkrittum19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-jitkrittum19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Wittawat
family: Jitkrittum
- given: Patsorn
family: Sangkloy
- given: Muhammad Waleed
family: Gondal
- given: Amit
family: Raj
- given: James
family: Hays
- given: Bernhard
family: Schölkopf
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3140-3151
id: jitkrittum19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3140
lastpage: 3151
published: 2019-05-24 00:00:00 +0000
- title: 'GOODE: A Gaussian Off-The-Shelf Ordinary Differential Equation Solver'
abstract: 'There are two types of ordinary differential equations (ODEs): initial value problems (IVPs) and boundary value problems (BVPs). While many probabilistic numerical methods for the solution of IVPs have been presented to-date, there exists no efficient probabilistic general-purpose solver for nonlinear BVPs. Our method based on iterated Gaussian process (GP) regression returns a GP posterior over the solution of nonlinear ODEs, which provides a meaningful error estimation via its predictive posterior standard deviation. Our solver is fast (typically of quadratic convergence rate) and the theory of convergence can be transferred from prior non-probabilistic work. Our method performs on par with standard codes for an established benchmark of test problems.'
volume: 97
URL: https://proceedings.mlr.press/v97/john19a.html
PDF: http://proceedings.mlr.press/v97/john19a/john19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-john19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: David
family: John
- given: Vincent
family: Heuveline
- given: Michael
family: Schober
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3152-3162
id: john19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3152
lastpage: 3162
published: 2019-05-24 00:00:00 +0000
- title: 'Bilinear Bandits with Low-rank Structure'
abstract: 'We introduce the bilinear bandit problem with low-rank structure in which an action takes the form of a pair of arms from two different entity types, and the reward is a bilinear function of the known feature vectors of the arms. The unknown in the problem is a $d_1$ by $d_2$ matrix $\mathbf{\Theta}^*$ that defines the reward, and has low rank $r \ll \min\{d_1,d_2\}$. Determination of $\mathbf{\Theta}^*$ with this low-rank structure poses a significant challenge in finding the right exploration-exploitation tradeoff. In this work, we propose a new two-stage algorithm called “Explore-Subspace-Then-Refine” (ESTR). The first stage is an explicit subspace exploration, while the second stage is a linear bandit algorithm called “almost-low-dimensional OFUL” (LowOFUL) that exploits and further refines the estimated subspace via a regularization technique. We show that the regret of ESTR is $\widetilde{\mathcal{O}}((d_1+d_2)^{3/2} \sqrt{r T})$ where $\widetilde{\mathcal{O}}$ hides logarithmic factors and $T$ is the time horizon, which improves upon the regret of $\widetilde{\mathcal{O}}(d_1d_2\sqrt{T})$ attained for a naïve linear bandit reduction. We conjecture that the regret bound of ESTR is unimprovable up to polylogarithmic factors, and our preliminary experiment shows that ESTR outperforms a naïve linear bandit reduction.'
volume: 97
URL: https://proceedings.mlr.press/v97/jun19a.html
PDF: http://proceedings.mlr.press/v97/jun19a/jun19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-jun19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Kwang-Sung
family: Jun
- given: Rebecca
family: Willett
- given: Stephen
family: Wright
- given: Robert
family: Nowak
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3163-3172
id: jun19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3163
lastpage: 3172
published: 2019-05-24 00:00:00 +0000
- title: 'Statistical Foundations of Virtual Democracy'
abstract: 'Virtual democracy is an approach to automating decisions, by learning models of the preferences of individual people, and, at runtime, aggregating the predicted preferences of those people on the dilemma at hand. One of the key questions is which aggregation method – or voting rule – to use; we offer a novel statistical viewpoint that provides guidance. Specifically, we seek voting rules that are robust to prediction errors, in that their output on people’s true preferences is likely to coincide with their output on noisy estimates thereof. We prove that the classic Borda count rule is robust in this sense, whereas any voting rule belonging to the wide family of pairwise-majority consistent rules is not. Our empirical results further support, and more precisely measure, the robustness of Borda count.'
volume: 97
URL: https://proceedings.mlr.press/v97/kahng19a.html
PDF: http://proceedings.mlr.press/v97/kahng19a/kahng19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kahng19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Anson
family: Kahng
- given: Min Kyung
family: Lee
- given: Ritesh
family: Noothigattu
- given: Ariel
family: Procaccia
- given: Christos-Alexandros
family: Psomas
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3173-3182
id: kahng19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3173
lastpage: 3182
published: 2019-05-24 00:00:00 +0000
- title: 'Molecular Hypergraph Grammar with Its Application to Molecular Optimization'
abstract: 'Molecular optimization aims to discover novel molecules with desirable properties, and its two fundamental challenges are: (i) it is not trivial to generate valid molecules in a controllable way due to hard chemical constraints such as the valency conditions, and (ii) it is often costly to evaluate a property of a novel molecule, and therefore, the number of property evaluations is limited. These challenges are to some extent alleviated by a combination of a variational autoencoder (VAE) and Bayesian optimization (BO), where VAE converts a molecule into/from its latent continuous vector, and BO optimizes a latent continuous vector (and its corresponding molecule) within a limited number of property evaluations. While the most recent work, for the first time, achieved 100% validity, its architecture is rather complex due to auxiliary neural networks other than VAE, making it difficult to train. This paper presents a molecular hypergraph grammar variational autoencoder (MHG-VAE), which uses a single VAE to achieve 100% validity. Our idea is to develop a graph grammar encoding the hard chemical constraints, called molecular hypergraph grammar (MHG), which guides VAE to always generate valid molecules. We also present an algorithm to construct MHG from a set of molecules.'
volume: 97
URL: https://proceedings.mlr.press/v97/kajino19a.html
PDF: http://proceedings.mlr.press/v97/kajino19a/kajino19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kajino19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Hiroshi
family: Kajino
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3183-3191
id: kajino19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3183
lastpage: 3191
published: 2019-05-24 00:00:00 +0000
- title: 'Robust Influence Maximization for Hyperparametric Models'
abstract: 'In this paper we study the problem of robust influence maximization in the independent cascade model under a hyperparametric assumption. In social networks users influence and are influenced by individuals with similar characteristics and as such they are associated with some features. A recent surging research direction in influence maximization focuses on the case where the edge probabilities on the graph are not arbitrary but are generated as a function of the features of the users and a global hyperparameter. We propose a model where the objective is to maximize the worst-case number of influenced users for any possible value of that hyperparameter. We provide theoretical results showing that proper robust solution in our model is NP-hard and an algorithm that achieves improper robust optimization. We make-use of sampling based techniques and of the renowned multiplicative weight updates algorithm. Additionally we validate our method empirically and prove that it outperforms the state-of-the-art robust influence maximization techniques.'
volume: 97
URL: https://proceedings.mlr.press/v97/kalimeris19a.html
PDF: http://proceedings.mlr.press/v97/kalimeris19a/kalimeris19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kalimeris19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Dimitris
family: Kalimeris
- given: Gal
family: Kaplun
- given: Yaron
family: Singer
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3192-3200
id: kalimeris19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3192
lastpage: 3200
published: 2019-05-24 00:00:00 +0000
- title: 'Classifying Treatment Responders Under Causal Effect Monotonicity'
abstract: 'In the context of individual-level causal inference, we study the problem of predicting whether someone will respond or not to a treatment based on their features and past examples of features, treatment indicator (e.g., drug/no drug), and a binary outcome (e.g., recovery from disease). As a classification task, the problem is made difficult by not knowing the example outcomes under the opposite treatment indicators. We assume the effect is monotonic, as in advertising’s effect on a purchase or bail-setting’s effect on reappearance in court: either it would have happened regardless of treatment, not happened regardless, or happened only depending on exposure to treatment. Predicting whether the latter is latently the case is our focus. While previous work focuses on conditional average treatment effect estimation, formulating the problem as a classification task allows us to develop new tools more suited to this problem. By leveraging monotonicity, we develop new discriminative and generative algorithms for the responder-classification problem. We explore and discuss connections to corrupted data and policy learning. We provide an empirical study with both synthetic and real datasets to compare these specialized algorithms to standard benchmarks.'
volume: 97
URL: https://proceedings.mlr.press/v97/kallus19a.html
PDF: http://proceedings.mlr.press/v97/kallus19a/kallus19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kallus19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Nathan
family: Kallus
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3201-3210
id: kallus19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3201
lastpage: 3210
published: 2019-05-24 00:00:00 +0000
- title: 'Trainable Decoding of Sets of Sequences for Neural Sequence Models'
abstract: 'Many sequence prediction tasks admit multiple correct outputs and so, it is often useful to decode a set of outputs that maximize some task-specific set-level metric. However, retooling standard sequence prediction procedures tailored towards predicting the single best output leads to the decoding of sets containing very similar sequences; failing to capture the variation in the output space. To address this, we propose $\nabla$BS, a trainable decoding procedure that outputs a set of sequences, highly valued according to the metric. Our method tightly integrates the training and decoding phases and further allows for the optimization of the task-specific metric addressing the shortcomings of standard sequence prediction. Further, we discuss the trade-offs of commonly used set-level metrics and motivate a new set-level metric that naturally evaluates the notion of “capturing the variation in the output space”. Finally, we show results on the image captioning task and find that our model outperforms standard techniques and natural ablations.'
volume: 97
URL: https://proceedings.mlr.press/v97/kalyan19a.html
PDF: http://proceedings.mlr.press/v97/kalyan19a/kalyan19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kalyan19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ashwin
family: Kalyan
- given: Peter
family: Anderson
- given: Stefan
family: Lee
- given: Dhruv
family: Batra
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3211-3221
id: kalyan19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3211
lastpage: 3221
published: 2019-05-24 00:00:00 +0000
- title: 'Myopic Posterior Sampling for Adaptive Goal Oriented Design of Experiments'
abstract: 'Bayesian methods for adaptive decision-making, such as Bayesian optimisation, active learning, and active search have seen great success in relevant applications. However, real world data collection tasks are more broad and complex, as we may need to achieve a combination of the above goals and/or application specific goals. In such scenarios, specialised methods have limited applicability. In this work, we design a new myopic strategy for a wide class of adaptive design of experiment (DOE) problems, where we wish to collect data in order to fulfil a given goal. Our approach, Myopic Posterior Sampling (MPS), which is inspired by the classical posterior sampling algorithm for multi-armed bandits, enables us to address a broad suite of DOE tasks where a practitioner may incorporate domain expertise about the system and specify her desired goal via a reward function. Empirically, this general-purpose strategy is competitive with more specialised methods in a wide array of synthetic and real world DOE tasks. More importantly, it enables addressing complex DOE goals where no existing method seems applicable. On the theoretical side, we leverage ideas from adaptive submodularity and reinforcement learning to derive conditions under which MPS achieves sublinear regret against natural benchmark policies.'
volume: 97
URL: https://proceedings.mlr.press/v97/kandasamy19a.html
PDF: http://proceedings.mlr.press/v97/kandasamy19a/kandasamy19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kandasamy19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Kirthevasan
family: Kandasamy
- given: Willie
family: Neiswanger
- given: Reed
family: Zhang
- given: Akshay
family: Krishnamurthy
- given: Jeff
family: Schneider
- given: Barnabas
family: Poczos
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3222-3232
id: kandasamy19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3222
lastpage: 3232
published: 2019-05-24 00:00:00 +0000
- title: 'Differentially Private Learning of Geometric Concepts'
abstract: 'We present differentially private efficient algorithms for learning union of polygons in the plane (which are not necessarily convex). Our algorithms achieve $(\alpha,\beta)$-PAC learning and $(\epsilon,\delta)$-differential privacy using a sample of size $\tilde{O}\left(\frac{1}{\alpha\epsilon}k\log d\right)$, where the domain is $[d]\times[d]$ and $k$ is the number of edges in the union of polygons.'
volume: 97
URL: https://proceedings.mlr.press/v97/kaplan19a.html
PDF: http://proceedings.mlr.press/v97/kaplan19a/kaplan19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kaplan19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Haim
family: Kaplan
- given: Yishay
family: Mansour
- given: Yossi
family: Matias
- given: Uri
family: Stemmer
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3233-3241
id: kaplan19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3233
lastpage: 3241
published: 2019-05-24 00:00:00 +0000
- title: 'Policy Consolidation for Continual Reinforcement Learning'
abstract: 'We propose a method for tackling catastrophic forgetting in deep reinforcement learning that is *agnostic* to the timescale of changes in the distribution of experiences, does not require knowledge of task boundaries and can adapt in *continuously* changing environments. In our *policy consolidation* model, the policy network interacts with a cascade of hidden networks that simultaneously remember the agent’s policy at a range of timescales and regularise the current policy by its own history, thereby improving its ability to learn without forgetting. We find that the model improves continual learning relative to baselines on a number of continuous control tasks in single-task, alternating two-task, and multi-agent competitive self-play settings.'
volume: 97
URL: https://proceedings.mlr.press/v97/kaplanis19a.html
PDF: http://proceedings.mlr.press/v97/kaplanis19a/kaplanis19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kaplanis19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Christos
family: Kaplanis
- given: Murray
family: Shanahan
- given: Claudia
family: Clopath
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3242-3251
id: kaplanis19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3242
lastpage: 3251
published: 2019-05-24 00:00:00 +0000
- title: 'Error Feedback Fixes SignSGD and other Gradient Compression Schemes'
abstract: 'Sign-based algorithms (e.g. signSGD) have been proposed as a biased gradient compression technique to alleviate the communication bottleneck in training large neural networks across multiple workers. We show simple convex counter-examples where signSGD does not converge to the optimum. Further, even when it does converge, signSGD may generalize poorly when compared with SGD. These issues arise because of the biased nature of the sign compression operator. We then show that using error-feedback, i.e. incorporating the error made by the compression operator into the next step, overcomes these issues. We prove that our algorithm (EF-SGD) with arbitrary compression operator achieves the same rate of convergence as SGD without any additional assumptions. Thus EF-SGD achieves gradient compression for free. Our experiments thoroughly substantiate the theory.'
volume: 97
URL: https://proceedings.mlr.press/v97/karimireddy19a.html
PDF: http://proceedings.mlr.press/v97/karimireddy19a/karimireddy19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-karimireddy19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sai Praneeth
family: Karimireddy
- given: Quentin
family: Rebjock
- given: Sebastian
family: Stich
- given: Martin
family: Jaggi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3252-3261
id: karimireddy19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3252
lastpage: 3261
published: 2019-05-24 00:00:00 +0000
- title: 'Riemannian adaptive stochastic gradient algorithms on matrix manifolds'
abstract: 'Adaptive stochastic gradient algorithms in the Euclidean space have attracted much attention lately. Such explorations on Riemannian manifolds, on the other hand, are relatively new, limited, and challenging. This is because of the intrinsic non-linear structure of the underlying manifold and the absence of a canonical coordinate system. In machine learning applications, however, most manifolds of interest are represented as matrices with notions of row and column subspaces. In addition, the implicit manifold-related constraints may also lie on such subspaces. For example, the Grassmann manifold is the set of column subspaces. To this end, such a rich structure should not be lost by transforming matrices to just a stack of vectors while developing optimization algorithms on manifolds. We propose novel stochastic gradient algorithms for problems on Riemannian matrix manifolds by adapting the row and column subspaces of gradients. Our algorithms are provably convergent and they achieve the convergence rate of order $O(log(T)/sqrt(T))$, where $T$ is the number of iterations. Our experiments illustrate that the proposed algorithms outperform existing Riemannian adaptive stochastic algorithms.'
volume: 97
URL: https://proceedings.mlr.press/v97/kasai19a.html
PDF: http://proceedings.mlr.press/v97/kasai19a/kasai19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kasai19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Hiroyuki
family: Kasai
- given: Pratik
family: Jawanpuria
- given: Bamdev
family: Mishra
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3262-3271
id: kasai19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3262
lastpage: 3271
published: 2019-05-24 00:00:00 +0000
- title: 'Neural Inverse Knitting: From Images to Manufacturing Instructions'
abstract: 'Motivated by the recent potential of mass customization brought by whole-garment knitting machines, we introduce the new problem of automatic machine instruction generation using a single image of the desired physical product, which we apply to machine knitting. We propose to tackle this problem by directly learning to synthesize regular machine instructions from real images. We create a cured dataset of real samples with their instruction counterpart and propose to use synthetic images to augment it in a novel way. We theoretically motivate our data mixing framework and show empirical results suggesting that making real images look more synthetic is beneficial in our problem setup.'
volume: 97
URL: https://proceedings.mlr.press/v97/kaspar19a.html
PDF: http://proceedings.mlr.press/v97/kaspar19a/kaspar19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kaspar19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alexandre
family: Kaspar
- given: Tae-Hyun
family: Oh
- given: Liane
family: Makatura
- given: Petr
family: Kellnhofer
- given: Wojciech
family: Matusik
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3272-3281
id: kaspar19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3272
lastpage: 3281
published: 2019-05-24 00:00:00 +0000
- title: 'Processing Megapixel Images with Deep Attention-Sampling Models'
abstract: 'Existing deep architectures cannot operate on very large signals such as megapixel images due to computational and memory constraints. To tackle this limitation, we propose a fully differentiable end-to-end trainable model that samples and processes only a fraction of the full resolution input image. The locations to process are sampled from an attention distribution computed from a low resolution view of the input. We refer to our method as attention sampling and it can process images of several megapixels with a standard single GPU setup. We show that sampling from the attention distribution results in an unbiased estimator of the full model with minimal variance, and we derive an unbiased estimator of the gradient that we use to train our model end-to-end with a normal SGD procedure. This new method is evaluated on three classification tasks, where we show that it allows to reduce computation and memory footprint by an order of magnitude for the same accuracy as classical architectures. We also show the consistency of the sampling that indeed focuses on informative parts of the input images.'
volume: 97
URL: https://proceedings.mlr.press/v97/katharopoulos19a.html
PDF: http://proceedings.mlr.press/v97/katharopoulos19a/katharopoulos19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-katharopoulos19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Angelos
family: Katharopoulos
- given: Francois
family: Fleuret
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3282-3291
id: katharopoulos19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3282
lastpage: 3291
published: 2019-05-24 00:00:00 +0000
- title: 'Robust Estimation of Tree Structured Gaussian Graphical Models'
abstract: 'Consider jointly Gaussian random variables whose conditional independence structure is specified by a graphical model. If we observe realizations of the variables, we can compute the covariance matrix, and it is well known that the support of the inverse covariance matrix corresponds to the edges of the graphical model. Instead, suppose we only have noisy observations. If the noise at each node is independent, we can compute the sum of the covariance matrix and an unknown diagonal. The inverse of this sum is (in general) dense. We ask: can the original independence structure be recovered? We address this question for tree structured graphical models. We prove that this problem is unidentifiable, but show that this unidentifiability is limited to a small class of candidate trees. We further present additional constraints under which the problem is identifiable. Finally, we provide an O(n^3) algorithm to find this equivalence class of trees.'
volume: 97
URL: https://proceedings.mlr.press/v97/katiyar19a.html
PDF: http://proceedings.mlr.press/v97/katiyar19a/katiyar19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-katiyar19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ashish
family: Katiyar
- given: Jessica
family: Hoffmann
- given: Constantine
family: Caramanis
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3292-3300
id: katiyar19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3292
lastpage: 3300
published: 2019-05-24 00:00:00 +0000
- title: 'Shallow-Deep Networks: Understanding and Mitigating Network Overthinking'
abstract: 'We characterize a prevalent weakness of deep neural networks (DNNs), ’overthinking’, which occurs when a DNN can reach correct predictions before its final layer. Overthinking is computationally wasteful, and it can also be destructive when, by the final layer, a correct prediction changes into a misclassification. Understanding overthinking requires studying how each prediction evolves during a DNN’s forward pass, which conventionally is opaque. For prediction transparency, we propose the Shallow-Deep Network (SDN), a generic modification to off-the-shelf DNNs that introduces internal classifiers. We apply SDN to four modern architectures, trained on three image classification tasks, to characterize the overthinking problem. We show that SDNs can mitigate the wasteful effect of overthinking with confidence-based early exits, which reduce the average inference cost by more than 50% and preserve the accuracy. We also find that the destructive effect occurs for 50% of misclassifications on natural inputs and that it can be induced, adversarially, with a recent backdooring attack. To mitigate this effect, we propose a new confusion metric to quantify the internal disagreements that will likely to lead to misclassifications.'
volume: 97
URL: https://proceedings.mlr.press/v97/kaya19a.html
PDF: http://proceedings.mlr.press/v97/kaya19a/kaya19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kaya19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yigitcan
family: Kaya
- given: Sanghyun
family: Hong
- given: Tudor
family: Dumitras
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3301-3310
id: kaya19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3301
lastpage: 3310
published: 2019-05-24 00:00:00 +0000
- title: 'Submodular Streaming in All Its Glory: Tight Approximation, Minimum Memory and Low Adaptive Complexity'
abstract: 'Streaming algorithms are generally judged by the quality of their solution, memory footprint, and computational complexity. In this paper, we study the problem of maximizing a monotone submodular function in the streaming setting with a cardinality constraint $k$. We first propose SIEVE-STREAMING++, which requires just one pass over the data, keeps only $O(k)$ elements and achieves the tight $\frac{1}{2}$-approximation guarantee. The best previously known streaming algorithms either achieve a suboptimal $\frac{1}{4}$-approximation with $\Theta(k)$ memory or the optimal $\frac{1}{2}$-approximation with $O(k\log k)$ memory. Next, we show that by buffering a small fraction of the stream and applying a careful filtering procedure, one can heavily reduce the number of adaptive computational rounds, thus substantially lowering the computational complexity of SIEVE-STREAMING++. We then generalize our results to the more challenging multi-source streaming setting. We show how one can achieve the tight $\frac{1}{2}$-approximation guarantee with $O(k)$ shared memory, while minimizing not only the rounds of computations but also the total number of communicated bits. Finally, we demonstrate the efficiency of our algorithms on real-world data summarization tasks for multi-source streams of tweets and of YouTube videos.'
volume: 97
URL: https://proceedings.mlr.press/v97/kazemi19a.html
PDF: http://proceedings.mlr.press/v97/kazemi19a/kazemi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kazemi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ehsan
family: Kazemi
- given: Marko
family: Mitrovic
- given: Morteza
family: Zadimoghaddam
- given: Silvio
family: Lattanzi
- given: Amin
family: Karbasi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3311-3320
id: kazemi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3311
lastpage: 3320
published: 2019-05-24 00:00:00 +0000
- title: 'Adaptive Scale-Invariant Online Algorithms for Learning Linear Models'
abstract: 'We consider online learning with linear models, where the algorithm predicts on sequentially revealed instances (feature vectors), and is compared against the best linear function (comparator) in hindsight. Popular algorithms in this framework, such as Online Gradient Descent (OGD), have parameters (learning rates), which ideally should be tuned based on the scales of the features and the optimal comparator, but these quantities only become available at the end of the learning process. In this paper, we resolve the tuning problem by proposing online algorithms making predictions which are invariant under arbitrary rescaling of the features. The algorithms have no parameters to tune, do not require any prior knowledge on the scale of the instances or the comparator, and achieve regret bounds matching (up to a logarithmic factor) that of OGD with optimally tuned separate learning rates per dimension, while retaining comparable runtime performance.'
volume: 97
URL: https://proceedings.mlr.press/v97/kempka19a.html
PDF: http://proceedings.mlr.press/v97/kempka19a/kempka19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kempka19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Michal
family: Kempka
- given: Wojciech
family: Kotlowski
- given: Manfred K.
family: Warmuth
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3321-3330
id: kempka19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3321
lastpage: 3330
published: 2019-05-24 00:00:00 +0000
- title: 'CHiVE: Varying Prosody in Speech Synthesis with a Linguistically Driven Dynamic Hierarchical Conditional Variational Network'
abstract: 'The prosodic aspects of speech signals produced by current text-to-speech systems are typically averaged over training material, and as such lack the variety and liveliness found in natural speech. To avoid monotony and averaged prosody contours, it is desirable to have a way of modeling the variation in the prosodic aspects of speech, so audio signals can be synthesized in multiple ways for a given text. We present a new, hierarchically structured conditional variational auto-encoder to generate prosodic features (fundamental frequency, energy and duration) suitable for use with a vocoder or a generative model like WaveNet. At inference time, an embedding representing the prosody of a sentence may be sampled from the variational layer to allow for prosodic variation. To efficiently capture the hierarchical nature of the linguistic input (words, syllables and phones), both the encoder and decoder parts of the auto-encoder are hierarchical, in line with the linguistic structure, with layers being clocked dynamically at the respective rates. We show in our experiments that our dynamic hierarchical network outperforms a non-hierarchical state-of-the-art baseline, and, additionally, that prosody transfer across sentences is possible by employing the prosody embedding of one sentence to generate the speech signal of another.'
volume: 97
URL: https://proceedings.mlr.press/v97/kenter19a.html
PDF: http://proceedings.mlr.press/v97/kenter19a/kenter19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kenter19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Tom
family: Kenter
- given: Vincent
family: Wan
- given: Chun-An
family: Chan
- given: Rob
family: Clark
- given: Jakub
family: Vit
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3331-3340
id: kenter19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3331
lastpage: 3340
published: 2019-05-24 00:00:00 +0000
- title: 'Collaborative Evolutionary Reinforcement Learning'
abstract: 'Deep reinforcement learning algorithms have been successfully applied to a range of challenging control tasks. However, these methods typically struggle with achieving effective exploration and are extremely sensitive to the choice of hyperparameters. One reason is that most approaches use a noisy version of their operating policy to explore - thereby limiting the range of exploration. In this paper, we introduce Collaborative Evolutionary Reinforcement Learning (CERL), a scalable framework that comprises a portfolio of policies that simultaneously explore and exploit diverse regions of the solution space. A collection of learners - typically proven algorithms like TD3 - optimize over varying time-horizons leading to this diverse portfolio. All learners contribute to and use a shared replay buffer to achieve greater sample efficiency. Computational resources are dynamically distributed to favor the best learners as a form of online algorithm selection. Neuroevolution binds this entire process to generate a single emergent learner that exceeds the capabilities of any individual learner. Experiments in a range of continuous control benchmarks demonstrate that the emergent learner significantly outperforms its composite learners while remaining overall more sample-efficient - notably solving the Mujoco Humanoid benchmark where all of its composite learners (TD3) fail entirely in isolation.'
volume: 97
URL: https://proceedings.mlr.press/v97/khadka19a.html
PDF: http://proceedings.mlr.press/v97/khadka19a/khadka19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-khadka19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Shauharda
family: Khadka
- given: Somdeb
family: Majumdar
- given: Tarek
family: Nassar
- given: Zach
family: Dwiel
- given: Evren
family: Tumer
- given: Santiago
family: Miret
- given: Yinyin
family: Liu
- given: Kagan
family: Tumer
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3341-3350
id: khadka19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3341
lastpage: 3350
published: 2019-05-24 00:00:00 +0000
- title: 'Geometry Aware Convolutional Filters for Omnidirectional Images Representation'
abstract: 'Due to their wide field of view, omnidirectional cameras are frequently used by autonomous vehicles, drones and robots for navigation and other computer vision tasks. The images captured by such cameras, are often analyzed and classified with techniques designed for planar images that unfortunately fail to properly handle the native geometry of such images and therefore results in suboptimal performance. In this paper we aim at improving popular deep convolutional neural networks so that they can properly take into account the specific properties of omnidirectional data. In particular we propose an algorithm that adapts convolutional layers, which often serve as a core building block of a CNN, to the properties of omnidirectional images. Thus, our filters have a shape and size that adapt to the location on the omnidirectional image. We show that our method is not limited to spherical surfaces and is able to incorporate the knowledge about any kind of projective geometry inside the deep learning network. As depicted by our experiments, our method outperforms the existing deep neural network techniques for omnidirectional image classification and compression tasks.'
volume: 97
URL: https://proceedings.mlr.press/v97/khasanova19a.html
PDF: http://proceedings.mlr.press/v97/khasanova19a/khasanova19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-khasanova19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Renata
family: Khasanova
- given: Pascal
family: Frossard
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3351-3359
id: khasanova19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3351
lastpage: 3359
published: 2019-05-24 00:00:00 +0000
- title: 'EMI: Exploration with Mutual Information'
abstract: 'Reinforcement learning algorithms struggle when the reward signal is very sparse. In these cases, naive random exploration methods essentially rely on a random walk to stumble onto a rewarding state. Recent works utilize intrinsic motivation to guide the exploration via generative models, predictive forward models, or discriminative modeling of novelty. We propose EMI, which is an exploration method that constructs embedding representation of states and actions that does not rely on generative decoding of the full observation but extracts predictive signals that can be used to guide exploration based on forward prediction in the representation space. Our experiments show competitive results on challenging locomotion tasks with continuous control and on image-based exploration tasks with discrete actions on Atari. The source code is available at https://github.com/snu-mllab/EMI.'
volume: 97
URL: https://proceedings.mlr.press/v97/kim19a.html
PDF: http://proceedings.mlr.press/v97/kim19a/kim19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kim19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Hyoungseok
family: Kim
- given: Jaekyeom
family: Kim
- given: Yeonwoo
family: Jeong
- given: Sergey
family: Levine
- given: Hyun Oh
family: Song
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3360-3369
id: kim19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3360
lastpage: 3369
published: 2019-05-24 00:00:00 +0000
- title: 'FloWaveNet : A Generative Flow for Raw Audio'
abstract: 'Most modern text-to-speech architectures use a WaveNet vocoder for synthesizing high-fidelity waveform audio, but there have been limitations, such as high inference time, in practical applications due to its ancestral sampling scheme. The recently suggested Parallel WaveNet and ClariNet has achieved real-time audio synthesis capability by incorporating inverse autoregressive flow (IAF) for parallel sampling. However, these approaches require a two-stage training pipeline with a well-trained teacher network and can only produce natural sound by using probability distillation along with heavily-engineered auxiliary loss terms. We propose FloWaveNet, a flow-based generative model for raw audio synthesis. FloWaveNet requires only a single-stage training procedure and a single maximum likelihood loss, without any additional auxiliary terms, and it is inherently parallel due to the characteristics of generative flow. The model can efficiently sample raw audio in real-time, with clarity comparable to previous two-stage parallel models. The code and samples for all models, including our FloWaveNet, are available on GitHub.'
volume: 97
URL: https://proceedings.mlr.press/v97/kim19b.html
PDF: http://proceedings.mlr.press/v97/kim19b/kim19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kim19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sungwon
family: Kim
- given: Sang-Gil
family: Lee
- given: Jongyoon
family: Song
- given: Jaehyeon
family: Kim
- given: Sungroh
family: Yoon
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3370-3378
id: kim19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3370
lastpage: 3378
published: 2019-05-24 00:00:00 +0000
- title: 'Curiosity-Bottleneck: Exploration By Distilling Task-Specific Novelty'
abstract: 'Exploration based on state novelty has brought great success in challenging reinforcement learning problems with sparse rewards. However, existing novelty-based strategies become inefficient in real-world problems where observation contains not only task-dependent state novelty of our interest but also task-irrelevant information that should be ignored. We introduce an information- theoretic exploration strategy named Curiosity-Bottleneck that distills task-relevant information from observation. Based on the information bottleneck principle, our exploration bonus is quantified as the compressiveness of observation with respect to the learned representation of a compressive value network. With extensive experiments on static image classification, grid-world and three hard-exploration Atari games, we show that Curiosity-Bottleneck learns an effective exploration strategy by robustly measuring the state novelty in distractive environments where state-of-the-art exploration methods often degenerate.'
volume: 97
URL: https://proceedings.mlr.press/v97/kim19c.html
PDF: http://proceedings.mlr.press/v97/kim19c/kim19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kim19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Youngjin
family: Kim
- given: Wontae
family: Nam
- given: Hyunwoo
family: Kim
- given: Ji-Hoon
family: Kim
- given: Gunhee
family: Kim
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3379-3388
id: kim19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3379
lastpage: 3388
published: 2019-05-24 00:00:00 +0000
- title: 'Contextual Multi-armed Bandit Algorithm for Semiparametric Reward Model'
abstract: 'Contextual multi-armed bandit (MAB) algorithms have been shown promising for maximizing cumulative rewards in sequential decision tasks such as news article recommendation systems, web page ad placement algorithms, and mobile health. However, most of the proposed contextual MAB algorithms assume linear relationships between the reward and the context of the action. This paper proposes a new contextual MAB algorithm for a relaxed, semiparametric reward model that supports nonstationarity. The proposed method is less restrictive, easier to implement and faster than two alternative algorithms that consider the same model, while achieving a tight regret upper bound. We prove that the high-probability upper bound of the regret incurred by the proposed algorithm has the same order as the Thompson sampling algorithm for linear reward models. The proposed and existing algorithms are evaluated via simulation and also applied to Yahoo! news article recommendation log data.'
volume: 97
URL: https://proceedings.mlr.press/v97/kim19d.html
PDF: http://proceedings.mlr.press/v97/kim19d/kim19d.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kim19d.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Gi-Soo
family: Kim
- given: Myunghee Cho
family: Paik
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3389-3397
id: kim19d
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3389
lastpage: 3397
published: 2019-05-24 00:00:00 +0000
- title: 'Uniform Convergence Rate of the Kernel Density Estimator Adaptive to Intrinsic Volume Dimension'
abstract: 'We derive concentration inequalities for the supremum norm of the difference between a kernel density estimator (KDE) and its point-wise expectation that hold uniformly over the selection of the bandwidth and under weaker conditions on the kernel and the data generating distribution than previously used in the literature. We first propose a novel concept, called the volume dimension, to measure the intrinsic dimension of the support of a probability distribution based on the rates of decay of the probability of vanishing Euclidean balls. Our bounds depend on the volume dimension and generalize the existing bounds derived in the literature. In particular, when the data-generating distribution has a bounded Lebesgue density or is supported on a sufficiently well-behaved lower-dimensional manifold, our bound recovers the same convergence rate depending on the intrinsic dimension of the support as ones known in the literature. At the same time, our results apply to more general cases, such as the ones of distribution with unbounded densities or supported on a mixture of manifolds with different dimensions. Analogous bounds are derived for the derivative of the KDE, of any order. Our results are generally applicable but are especially useful for problems in geometric inference and topological data analysis, including level set estimation, density-based clustering, modal clustering and mode hunting, ridge estimation and persistent homology.'
volume: 97
URL: https://proceedings.mlr.press/v97/kim19e.html
PDF: http://proceedings.mlr.press/v97/kim19e/kim19e.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kim19e.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jisu
family: Kim
- given: Jaehyeok
family: Shin
- given: Alessandro
family: Rinaldo
- given: Larry
family: Wasserman
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3398-3407
id: kim19e
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3398
lastpage: 3407
published: 2019-05-24 00:00:00 +0000
- title: 'Bit-Swap: Recursive Bits-Back Coding for Lossless Compression with Hierarchical Latent Variables'
abstract: 'The bits-back argument suggests that latent variable models can be turned into lossless compression schemes. Translating the bits-back argument into efficient and practical lossless compression schemes for general latent variable models, however, is still an open problem. Bits-Back with Asymmetric Numeral Systems (BB-ANS), recently proposed by Townsend et al,. 2019, makes bits-back coding practically feasible for latent variable models with one latent layer, but it is inefficient for hierarchical latent variable models. In this paper we propose Bit-Swap, a new compression scheme that generalizes BB-ANS and achieves strictly better compression rates for hierarchical latent variable models with Markov chain structure. Through experiments we verify that Bit-Swap results in lossless compression rates that are empirically superior to existing techniques.'
volume: 97
URL: https://proceedings.mlr.press/v97/kingma19a.html
PDF: http://proceedings.mlr.press/v97/kingma19a/kingma19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kingma19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Friso
family: Kingma
- given: Pieter
family: Abbeel
- given: Jonathan
family: Ho
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3408-3417
id: kingma19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3408
lastpage: 3417
published: 2019-05-24 00:00:00 +0000
- title: 'CompILE: Compositional Imitation Learning and Execution'
abstract: 'We introduce Compositional Imitation Learning and Execution (CompILE): a framework for learning reusable, variable-length segments of hierarchically-structured behavior from demonstration data. CompILE uses a novel unsupervised, fully-differentiable sequence segmentation module to learn latent encodings of sequential data that can be re-composed and executed to perform new tasks. Once trained, our model generalizes to sequences of longer length and from environment instances not seen during training. We evaluate CompILE in a challenging 2D multi-task environment and a continuous control task, and show that it can find correct task boundaries and event encodings in an unsupervised manner. Latent codes and associated behavior policies discovered by CompILE can be used by a hierarchical agent, where the high-level policy selects actions in the latent code space, and the low-level, task-specific policies are simply the learned decoders. We found that our CompILE-based agent could learn given only sparse rewards, where agents without task-specific policies struggle.'
volume: 97
URL: https://proceedings.mlr.press/v97/kipf19a.html
PDF: http://proceedings.mlr.press/v97/kipf19a/kipf19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kipf19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Thomas
family: Kipf
- given: Yujia
family: Li
- given: Hanjun
family: Dai
- given: Vinicius
family: Zambaldi
- given: Alvaro
family: Sanchez-Gonzalez
- given: Edward
family: Grefenstette
- given: Pushmeet
family: Kohli
- given: Peter
family: Battaglia
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3418-3428
id: kipf19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3418
lastpage: 3428
published: 2019-05-24 00:00:00 +0000
- title: 'Adaptive and Safe Bayesian Optimization in High Dimensions via One-Dimensional Subspaces'
abstract: 'Bayesian optimization is known to be difficult to scale to high dimensions, because the acquisition step requires solving a non-convex optimization problem in the same search space. In order to scale the method and keep its benefits, we propose an algorithm (LineBO) that restricts the problem to a sequence of iteratively chosen one-dimensional sub-problems that can be solved efficiently. We show that our algorithm converges globally and obtains a fast local rate when the function is strongly convex. Further, if the objective has an invariant subspace, our method automatically adapts to the effective dimension without changing the algorithm. When combined with the SafeOpt algorithm to solve the sub-problems, we obtain the first safe Bayesian optimization algorithm with theoretical guarantees applicable in high-dimensional settings. We evaluate our method on multiple synthetic benchmarks, where we obtain competitive performance. Further, we deploy our algorithm to optimize the beam intensity of the Swiss Free Electron Laser with up to 40 parameters while satisfying safe operation constraints.'
volume: 97
URL: https://proceedings.mlr.press/v97/kirschner19a.html
PDF: http://proceedings.mlr.press/v97/kirschner19a/kirschner19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kirschner19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Johannes
family: Kirschner
- given: Mojmir
family: Mutny
- given: Nicole
family: Hiller
- given: Rasmus
family: Ischebeck
- given: Andreas
family: Krause
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3429-3438
id: kirschner19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3429
lastpage: 3438
published: 2019-05-24 00:00:00 +0000
- title: 'AUCμ: A Performance Metric for Multi-Class Machine Learning Models'
abstract: 'The area under the receiver operating characteristic curve (AUC) is arguably the most common metric in machine learning for assessing the quality of a two-class classification model. As the number and complexity of machine learning applications grows, so too does the need for measures that can gracefully extend to classification models trained for more than two classes. Prior work in this area has proven computationally intractable and/or inconsistent with known properties of AUC, and thus there is still a need for an improved multi-class efficacy metric. We provide in this work a multi-class extension of AUC that we call AUC{\textmu} that is derived from first principles of the binary class AUC. AUC{\textmu} has similar computational complexity to AUC and maintains the properties of AUC critical to its interpretation and use.'
volume: 97
URL: https://proceedings.mlr.press/v97/kleiman19a.html
PDF: http://proceedings.mlr.press/v97/kleiman19a/kleiman19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kleiman19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ross
family: Kleiman
- given: David
family: Page
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3439-3447
id: kleiman19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3439
lastpage: 3447
published: 2019-05-24 00:00:00 +0000
- title: 'Fair k-Center Clustering for Data Summarization'
abstract: 'In data summarization we want to choose $k$ prototypes in order to summarize a data set. We study a setting where the data set comprises several demographic groups and we are restricted to choose $k_i$ prototypes belonging to group $i$. A common approach to the problem without the fairness constraint is to optimize a centroid-based clustering objective such as $k$-center. A natural extension then is to incorporate the fairness constraint into the clustering problem. Existing algorithms for doing so run in time super-quadratic in the size of the data set, which is in contrast to the standard $k$-center problem being approximable in linear time. In this paper, we resolve this gap by providing a simple approximation algorithm for the $k$-center problem under the fairness constraint with running time linear in the size of the data set and $k$. If the number of demographic groups is small, the approximation guarantee of our algorithm only incurs a constant-factor overhead.'
volume: 97
URL: https://proceedings.mlr.press/v97/kleindessner19a.html
PDF: http://proceedings.mlr.press/v97/kleindessner19a/kleindessner19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kleindessner19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Matthäus
family: Kleindessner
- given: Pranjal
family: Awasthi
- given: Jamie
family: Morgenstern
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3448-3457
id: kleindessner19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3448
lastpage: 3457
published: 2019-05-24 00:00:00 +0000
- title: 'Guarantees for Spectral Clustering with Fairness Constraints'
abstract: 'Given the widespread popularity of spectral clustering (SC) for partitioning graph data, we study a version of constrained SC in which we try to incorporate the fairness notion proposed by Chierichetti et al. (2017). According to this notion, a clustering is fair if every demographic group is approximately proportionally represented in each cluster. To this end, we develop variants of both normalized and unnormalized constrained SC and show that they help find fairer clusterings on both synthetic and real data. We also provide a rigorous theoretical analysis of our algorithms on a natural variant of the stochastic block model, where $h$ groups have strong inter-group connectivity, but also exhibit a “natural” clustering structure which is fair. We prove that our algorithms can recover this fair clustering with high probability.'
volume: 97
URL: https://proceedings.mlr.press/v97/kleindessner19b.html
PDF: http://proceedings.mlr.press/v97/kleindessner19b/kleindessner19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kleindessner19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Matthäus
family: Kleindessner
- given: Samira
family: Samadi
- given: Pranjal
family: Awasthi
- given: Jamie
family: Morgenstern
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3458-3467
id: kleindessner19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3458
lastpage: 3467
published: 2019-05-24 00:00:00 +0000
- title: 'POPQORN: Quantifying Robustness of Recurrent Neural Networks'
abstract: 'The vulnerability to adversarial attacks has been a critical issue for deep neural networks. Addressing this issue requires a reliable way to evaluate the robustness of a network. Recently, several methods have been developed to compute robustness quantification for neural networks, namely, certified lower bounds of the minimum adversarial perturbation. Such methods, however, were devised for feed-forward networks, e.g. multi-layer perceptron or convolutional networks. It remains an open problem to quantify robustness for recurrent networks, especially LSTM and GRU. For such networks, there exist additional challenges in computing the robustness quantification, such as handling the inputs at multiple steps and the interaction between gates and states. In this work, we propose POPQORN (Propagated-output Quantified Robustness for RNNs), a general algorithm to quantify robustness of RNNs, including vanilla RNNs, LSTMs, and GRUs. We demonstrate its effectiveness on different network architectures and show that the robustness quantification on individual steps can lead to new insights.'
volume: 97
URL: https://proceedings.mlr.press/v97/ko19a.html
PDF: http://proceedings.mlr.press/v97/ko19a/ko19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ko19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ching-Yun
family: Ko
- given: Zhaoyang
family: Lyu
- given: Lily
family: Weng
- given: Luca
family: Daniel
- given: Ngai
family: Wong
- given: Dahua
family: Lin
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3468-3477
id: ko19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3468
lastpage: 3477
published: 2019-05-24 00:00:00 +0000
- title: 'Decentralized Stochastic Optimization and Gossip Algorithms with Compressed Communication'
abstract: 'We consider decentralized stochastic optimization with the objective function (e.g. data samples for machine learning tasks) being distributed over n machines that can only communicate to their neighbors on a fixed communication graph. To address the communication bottleneck, the nodes compress (e.g. quantize or sparsify) their model updates. We cover both unbiased and biased compression operators with quality denoted by \delta <= 1 (\delta=1 meaning no compression). We (i) propose a novel gossip-based stochastic gradient descent algorithm, CHOCO-SGD, that converges at rate O(1/(nT) + 1/(T \rho^2 \delta)^2) for strongly convex objectives, where T denotes the number of iterations and \rho the eigengap of the connectivity matrix. We (ii) present a novel gossip algorithm, CHOCO-GOSSIP, for the average consensus problem that converges in time O(1/(\rho^2\delta) \log (1/\epsilon)) for accuracy \epsilon > 0. This is (up to our knowledge) the first gossip algorithm that supports arbitrary compressed messages for \delta > 0 and still exhibits linear convergence. We (iii) show in experiments that both of our algorithms do outperform the respective state-of-the-art baselines and CHOCO-SGD can reduce communication by at least two orders of magnitudes.'
volume: 97
URL: https://proceedings.mlr.press/v97/koloskova19a.html
PDF: http://proceedings.mlr.press/v97/koloskova19a/koloskova19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-koloskova19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Anastasia
family: Koloskova
- given: Sebastian
family: Stich
- given: Martin
family: Jaggi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3478-3487
id: koloskova19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3478
lastpage: 3487
published: 2019-05-24 00:00:00 +0000
- title: 'Robust Learning from Untrusted Sources'
abstract: 'Modern machine learning methods often require more data for training than a single expert can provide. Therefore, it has become a standard procedure to collect data from multiple external sources, \eg via crowdsourcing. Unfortunately, the quality of these sources is not always guaranteed. As further complications, the data might be stored in a distributed way, or might even have to remain private. In this work, we address the question of how to learn robustly in such scenarios. Studying the problem through the lens of statistical learning theory, we derive a procedure that allows for learning from all available sources, yet automatically suppresses irrelevant or corrupted data. We show by extensive experiments that our method provides significant improvements over alternative approaches from robust statistics and distributed optimization.'
volume: 97
URL: https://proceedings.mlr.press/v97/konstantinov19a.html
PDF: http://proceedings.mlr.press/v97/konstantinov19a/konstantinov19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-konstantinov19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Nikola
family: Konstantinov
- given: Christoph
family: Lampert
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3488-3498
id: konstantinov19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3488
lastpage: 3498
published: 2019-05-24 00:00:00 +0000
- title: 'Stochastic Beams and Where To Find Them: The Gumbel-Top-k Trick for Sampling Sequences Without Replacement'
abstract: 'The well-known Gumbel-Max trick for sampling from a categorical distribution can be extended to sample $k$ elements without replacement. We show how to implicitly apply this ’Gumbel-Top-$k$’ trick on a factorized distribution over sequences, allowing to draw exact samples without replacement using a Stochastic Beam Search. Even for exponentially large domains, the number of model evaluations grows only linear in $k$ and the maximum sampled sequence length. The algorithm creates a theoretical connection between sampling and (deterministic) beam search and can be used as a principled intermediate alternative. In a translation task, the proposed method compares favourably against alternatives to obtain diverse yet good quality translations. We show that sequences sampled without replacement can be used to construct low-variance estimators for expected sentence-level BLEU score and model entropy.'
volume: 97
URL: https://proceedings.mlr.press/v97/kool19a.html
PDF: http://proceedings.mlr.press/v97/kool19a/kool19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kool19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Wouter
family: Kool
- given: Herke
family: Van Hoof
- given: Max
family: Welling
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3499-3508
id: kool19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3499
lastpage: 3508
published: 2019-05-24 00:00:00 +0000
- title: 'LIT: Learned Intermediate Representation Training for Model Compression'
abstract: 'Researchers have proposed a range of model compression techniques to reduce the computational and memory footprint of deep neural networks (DNNs). In this work, we introduce Learned Intermediate representation Training (LIT), a novel model compression technique that outperforms a range of recent model compression techniques by leveraging the highly repetitive structure of modern DNNs (e.g., ResNet). LIT uses a teacher DNN to train a student DNN of reduced depth by leveraging two key ideas: 1) LIT directly compares intermediate representations of the teacher and student model and 2) LIT uses the intermediate representation from the teacher model’s previous block as input to the current student block during training, improving stability of intermediate representations in the student network. We show that LIT can substantially reduce network size without loss in accuracy on a range of DNN architectures and datasets. For example, LIT can compress ResNet on CIFAR10 by 3.4$\times$ outperforming network slimming and FitNets. Furthermore, LIT can compress, by depth, ResNeXt 5.5$\times$ on CIFAR10 (image classification), VDCNN by 1.7$\times$ on Amazon Reviews (sentiment analysis), and StarGAN by 1.8$\times$ on CelebA (style transfer, i.e., GANs).'
volume: 97
URL: https://proceedings.mlr.press/v97/koratana19a.html
PDF: http://proceedings.mlr.press/v97/koratana19a/koratana19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-koratana19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Animesh
family: Koratana
- given: Daniel
family: Kang
- given: Peter
family: Bailis
- given: Matei
family: Zaharia
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3509-3518
id: koratana19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3509
lastpage: 3518
published: 2019-05-24 00:00:00 +0000
- title: 'Similarity of Neural Network Representations Revisited'
abstract: 'Recent work has sought to understand the behavior of neural networks by comparing representations between layers and between different trained models. We examine methods for comparing neural network representations based on canonical correlation analysis (CCA). We show that CCA belongs to a family of statistics for measuring multivariate similarity, but that neither CCA nor any other statistic that is invariant to invertible linear transformation can measure meaningful similarities between representations of higher dimension than the number of data points. We introduce a similarity index that measures the relationship between representational similarity matrices and does not suffer from this limitation. This similarity index is equivalent to centered kernel alignment (CKA) and is also closely connected to CCA. Unlike CCA, CKA can reliably identify correspondences between representations in networks trained from different initializations.'
volume: 97
URL: https://proceedings.mlr.press/v97/kornblith19a.html
PDF: http://proceedings.mlr.press/v97/kornblith19a/kornblith19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kornblith19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Simon
family: Kornblith
- given: Mohammad
family: Norouzi
- given: Honglak
family: Lee
- given: Geoffrey
family: Hinton
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3519-3529
id: kornblith19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3519
lastpage: 3529
published: 2019-05-24 00:00:00 +0000
- title: 'On the Complexity of Approximating Wasserstein Barycenters'
abstract: 'We study the complexity of approximating the Wasserstein barycenter of $m$ discrete measures, or histograms of size $n$, by contrasting two alternative approaches that use entropic regularization. The first approach is based on the Iterative Bregman Projections (IBP) algorithm for which our novel analysis gives a complexity bound proportional to ${mn^2}/{\varepsilon^2}$ to approximate the original non-regularized barycenter. On the other hand, using an approach based on accelerated gradient descent, we obtain a complexity proportional to ${mn^{2}}/{\varepsilon}$. As a byproduct, we show that the regularization parameter in both approaches has to be proportional to $\varepsilon$, which causes instability of both algorithms when the desired accuracy is high. To overcome this issue, we propose a novel proximal-IBP algorithm, which can be seen as a proximal gradient method, which uses IBP on each iteration to make a proximal step. We also consider the question of scalability of these algorithms using approaches from distributed optimization and show that the first algorithm can be implemented in a centralized distributed setting (master/slave), while the second one is amenable to a more general decentralized distributed setting with an arbitrary network topology.'
volume: 97
URL: https://proceedings.mlr.press/v97/kroshnin19a.html
PDF: http://proceedings.mlr.press/v97/kroshnin19a/kroshnin19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kroshnin19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alexey
family: Kroshnin
- given: Nazarii
family: Tupitsa
- given: Darina
family: Dvinskikh
- given: Pavel
family: Dvurechensky
- given: Alexander
family: Gasnikov
- given: Cesar
family: Uribe
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3530-3540
id: kroshnin19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3530
lastpage: 3540
published: 2019-05-24 00:00:00 +0000
- title: 'Estimate Sequences for Variance-Reduced Stochastic Composite Optimization'
abstract: 'In this paper, we propose a unified view of gradient-based algorithms for stochastic convex composite optimization by extending the concept of estimate sequence introduced by Nesterov. This point of view covers the stochastic gradient descent method, variants of the approaches SAGA, SVRG, and has several advantages: (i) we provide a generic proof of convergence for the aforementioned methods; (ii) we show that this SVRG variant is adaptive to strong convexity; (iii) we naturally obtain new algorithms with the same guarantees; (iv) we derive generic strategies to make these algorithms robust to stochastic noise, which is useful when data is corrupted by small random perturbations. Finally, we show that this viewpoint is useful to obtain new accelerated algorithms in the sense of Nesterov.'
volume: 97
URL: https://proceedings.mlr.press/v97/kulunchakov19a.html
PDF: http://proceedings.mlr.press/v97/kulunchakov19a/kulunchakov19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kulunchakov19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Andrei
family: Kulunchakov
- given: Julien
family: Mairal
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3541-3550
id: kulunchakov19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3541
lastpage: 3550
published: 2019-05-24 00:00:00 +0000
- title: 'Faster Algorithms for Binary Matrix Factorization'
abstract: 'We give faster approximation algorithms for well-studied variants of Binary Matrix Factorization (BMF), where we are given a binary $m \times n$ matrix $A$ and would like to find binary rank-$k$ matrices $U, V$ to minimize the Frobenius norm of $U \cdot V - A$. In the first setting, $U \cdot V$ denotes multiplication over $\mathbb{Z}$, and we give a constant-factor approximation algorithm that runs in $2^{O(k^2 \log k)} \textrm{poly}(mn)$ time, improving upon the previous $\min(2^{2^k}, 2^n) \textrm{poly}(mn)$ time. Our techniques generalize to minimizing $\|U \cdot V - A\|_p$ for $p \geq 1$, in $2^{O(k^{\lceil p/2 \rceil + 1}\log k)} \textrm{poly}(mn)$ time. For $p = 1$, this has a graph-theoretic consequence, namely, a $2^{O(k^2)} \poly(mn)$-time algorithm to approximate a graph as a union of disjoint bicliques. In the second setting, $U \cdot V$ is over $\GF(2)$, and we give a bicriteria constant-factor approximation algorithm that runs in $2^{O(k^3)} \poly(mn)$ time to find binary rank-$O(k \log m)$ matrices $U$, $V$ whose cost is as good as the best rank-$k$ approximation, improving upon $\min(2^{2^k}mn, \min(m,n)^{k^{O(1)}} \textrm{poly}(mn))$ time.'
volume: 97
URL: https://proceedings.mlr.press/v97/kumar19a.html
PDF: http://proceedings.mlr.press/v97/kumar19a/kumar19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kumar19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ravi
family: Kumar
- given: Rina
family: Panigrahy
- given: Ali
family: Rahimi
- given: David
family: Woodruff
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3551-3559
id: kumar19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3551
lastpage: 3559
published: 2019-05-24 00:00:00 +0000
- title: 'Loss Landscapes of Regularized Linear Autoencoders'
abstract: 'Autoencoders are a deep learning model for representation learning. When trained to minimize the distance between the data and its reconstruction, linear autoencoders (LAEs) learn the subspace spanned by the top principal directions but cannot learn the principal directions themselves. In this paper, we prove that $L_2$-regularized LAEs are symmetric at all critical points and learn the principal directions as the left singular vectors of the decoder. We smoothly parameterize the critical manifold and relate the minima to the MAP estimate of probabilistic PCA. We illustrate these results empirically and consider implications for PCA algorithms, computational neuroscience, and the algebraic topology of learning.'
volume: 97
URL: https://proceedings.mlr.press/v97/kunin19a.html
PDF: http://proceedings.mlr.press/v97/kunin19a/kunin19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kunin19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Daniel
family: Kunin
- given: Jonathan
family: Bloom
- given: Aleksandrina
family: Goeva
- given: Cotton
family: Seed
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3560-3569
id: kunin19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3560
lastpage: 3569
published: 2019-05-24 00:00:00 +0000
- title: 'Geometry and Symmetry in Short-and-Sparse Deconvolution'
abstract: 'We study the Short-and-Sparse (SaS) deconvolution problem of recovering a short signal a0 and a sparse signal x0 from their convolution. We propose a method based on nonconvex optimization, which under certain conditions recovers the target short and sparse signals, up to a signed shift symmetry which is intrinsic to this model. This symmetry plays a central role in shaping the optimization landscape for deconvolution. We give a regional analysis, which characterizes this landscape geometrically, on a union of subspaces. Our geometric characterization holds when the length-p0 short signal a0 has shift coherence {\textmu}, and x0 follows a random sparsity model with sparsity rate $\theta$ $\in$ [c1/p0, c2/(p0\sqrt{\mu}+\sqrt{p0})] / (log^2(p0)) . Based on this geometry, we give a provable method that successfully solves SaS deconvolution with high probability.'
volume: 97
URL: https://proceedings.mlr.press/v97/kuo19a.html
PDF: http://proceedings.mlr.press/v97/kuo19a/kuo19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kuo19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Han-Wen
family: Kuo
- given: Yenson
family: Lau
- given: Yuqian
family: Zhang
- given: John
family: Wright
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3570-3580
id: kuo19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3570
lastpage: 3580
published: 2019-05-24 00:00:00 +0000
- title: 'A Large-Scale Study on Regularization and Normalization in GANs'
abstract: 'Generative adversarial networks (GANs) are a class of deep generative models which aim to learn a target distribution in an unsupervised fashion. While they were successfully applied to many problems, training a GAN is a notoriously challenging task and requires a significant number of hyperparameter tuning, neural architecture engineering, and a non-trivial amount of “tricks". The success in many practical applications coupled with the lack of a measure to quantify the failure modes of GANs resulted in a plethora of proposed losses, regularization and normalization schemes, as well as neural architectures. In this work we take a sober view of the current state of GANs from a practical perspective. We discuss and evaluate common pitfalls and reproducibility issues, open-source our code on Github, and provide pre-trained models on TensorFlow Hub.'
volume: 97
URL: https://proceedings.mlr.press/v97/kurach19a.html
PDF: http://proceedings.mlr.press/v97/kurach19a/kurach19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kurach19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Karol
family: Kurach
- given: Mario
family: Lučić
- given: Xiaohua
family: Zhai
- given: Marcin
family: Michalski
- given: Sylvain
family: Gelly
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3581-3590
id: kurach19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3581
lastpage: 3590
published: 2019-05-24 00:00:00 +0000
- title: 'Making Decisions that Reduce Discriminatory Impacts'
abstract: 'As machine learning algorithms move into real-world settings, it is crucial to ensure they are aligned with societal values. There has been much work on one aspect of this, namely the discriminatory prediction problem: How can we reduce discrimination in the predictions themselves? While an important question, solutions to this problem only apply in a restricted setting, as we have full control over the predictions. Often we care about the non-discrimination of quantities we do not have full control over. Thus, we describe another key aspect of this challenge, the discriminatory impact problem: How can we reduce discrimination arising from the real-world impact of decisions? To address this, we describe causal methods that model the relevant parts of the real-world system in which the decisions are made. Unlike previous approaches, these models not only allow us to map the causal pathway of a single decision, but also to model the effect of interference–how the impact on an individual depends on decisions made about other people. Often, the goal of decision policies is to maximize a beneficial impact overall. To reduce the discrimination of these benefits, we devise a constraint inspired by recent work in counterfactual fairness, and give an efficient procedure to solve the constrained optimization problem. We demonstrate our approach with an example: how to increase students taking college entrance exams in New York City public schools.'
volume: 97
URL: https://proceedings.mlr.press/v97/kusner19a.html
PDF: http://proceedings.mlr.press/v97/kusner19a/kusner19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kusner19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Matt
family: Kusner
- given: Chris
family: Russell
- given: Joshua
family: Loftus
- given: Ricardo
family: Silva
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3591-3600
id: kusner19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3591
lastpage: 3600
published: 2019-05-24 00:00:00 +0000
- title: 'Garbage In, Reward Out: Bootstrapping Exploration in Multi-Armed Bandits'
abstract: 'We propose a bandit algorithm that explores by randomizing its history of rewards. Specifically, it pulls the arm with the highest mean reward in a non-parametric bootstrap sample of its history with pseudo rewards. We design the pseudo rewards such that the bootstrap mean is optimistic with a sufficiently high probability. We call our algorithm Giro, which stands for garbage in, reward out. We analyze Giro in a Bernoulli bandit and derive a $O(K \Delta^{-1} \log n)$ bound on its $n$-round regret, where $\Delta$ is the difference in the expected rewards of the optimal and the best suboptimal arms, and $K$ is the number of arms. The main advantage of our exploration design is that it easily generalizes to structured problems. To show this, we propose contextual Giro with an arbitrary reward generalization model. We evaluate Giro and its contextual variant on multiple synthetic and real-world problems, and observe that it performs well.'
volume: 97
URL: https://proceedings.mlr.press/v97/kveton19a.html
PDF: http://proceedings.mlr.press/v97/kveton19a/kveton19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-kveton19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Branislav
family: Kveton
- given: Csaba
family: Szepesvari
- given: Sharan
family: Vaswani
- given: Zheng
family: Wen
- given: Tor
family: Lattimore
- given: Mohammad
family: Ghavamzadeh
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3601-3610
id: kveton19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3601
lastpage: 3610
published: 2019-05-24 00:00:00 +0000
- title: 'Characterizing Well-Behaved vs. Pathological Deep Neural Networks'
abstract: 'We introduce a novel approach, requiring only mild assumptions, for the characterization of deep neural networks at initialization. Our approach applies both to fully-connected and convolutional networks and easily incorporates batch normalization and skip-connections. Our key insight is to consider the evolution with depth of statistical moments of signal and noise, thereby characterizing the presence or absence of pathologies in the hypothesis space encoded by the choice of hyperparameters. We establish: (i) for feedforward networks, with and without batch normalization, the multiplicativity of layer composition inevitably leads to ill-behaved moments and pathologies; (ii) for residual networks with batch normalization, on the other hand, skip-connections induce power-law rather than exponential behaviour, leading to well-behaved moments and no pathology.'
volume: 97
URL: https://proceedings.mlr.press/v97/labatie19a.html
PDF: http://proceedings.mlr.press/v97/labatie19a/labatie19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-labatie19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Antoine
family: Labatie
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3611-3621
id: labatie19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3611
lastpage: 3621
published: 2019-05-24 00:00:00 +0000
- title: 'State-Reification Networks: Improving Generalization by Modeling the Distribution of Hidden Representations'
abstract: 'Machine learning promises methods that generalize well from finite labeled data. However, the brittleness of existing neural net approaches is revealed by notable failures, such as the existence of adversarial examples that are misclassified despite being nearly identical to a training example, or the inability of recurrent sequence-processing nets to stay on track without teacher forcing. We introduce a method, which we refer to as _state reification_, that involves modeling the distribution of hidden states over the training data and then projecting hidden states observed during testing toward this distribution. Our intuition is that if the network can remain in a familiar manifold of hidden space, subsequent layers of the net should be well trained to respond appropriately. We show that this state-reification method helps neural nets to generalize better, especially when labeled data are sparse, and also helps overcome the challenge of achieving robust generalization with adversarial training.'
volume: 97
URL: https://proceedings.mlr.press/v97/lamb19a.html
PDF: http://proceedings.mlr.press/v97/lamb19a/lamb19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lamb19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alex
family: Lamb
- given: Jonathan
family: Binas
- given: Anirudh
family: Goyal
- given: Sandeep
family: Subramanian
- given: Ioannis
family: Mitliagkas
- given: Yoshua
family: Bengio
- given: Michael
family: Mozer
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3622-3631
id: lamb19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3622
lastpage: 3631
published: 2019-05-24 00:00:00 +0000
- title: 'A Recurrent Neural Cascade-based Model for Continuous-Time Diffusion'
abstract: 'Many works have been proposed in the literature to capture the dynamics of diffusion in networks. While some of them define graphical Markovian models to extract temporal relationships between node infections in networks, others consider diffusion episodes as sequences of infections via recurrent neural models. In this paper we propose a model at the crossroads of these two extremes, which embeds the history of diffusion in infected nodes as hidden continuous states. Depending on the trajectory followed by the content before reaching a given node, the distribution of influence probabilities may vary. However, content trajectories are usually hidden in the data, which induces challenging learning problems. We propose a topological recurrent neural model which exhibits good experimental performances for diffusion modeling and prediction.'
volume: 97
URL: https://proceedings.mlr.press/v97/lamprier19a.html
PDF: http://proceedings.mlr.press/v97/lamprier19a/lamprier19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lamprier19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sylvain
family: Lamprier
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3632-3641
id: lamprier19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3632
lastpage: 3641
published: 2019-05-24 00:00:00 +0000
- title: 'Projection onto Minkowski Sums with Application to Constrained Learning'
abstract: 'We introduce block descent algorithms for projecting onto Minkowski sums of sets. Projection onto such sets is a crucial step in many statistical learning problems, and may regularize complexity of solutions to an optimization problem or arise in dual formulations of penalty methods. We show that projecting onto the Minkowski sum admits simple, efficient algorithms when complications such as overlapping constraints pose challenges to existing methods. We prove that our algorithm converges linearly when sets are strongly convex or satisfy an error bound condition, and extend the theory and methods to encompass non-convex sets as well. We demonstrate empirical advantages in runtime and accuracy over competitors in applications to $\ell_{1,p}$-regularized learning, constrained lasso, and overlapping group lasso.'
volume: 97
URL: https://proceedings.mlr.press/v97/lange19a.html
PDF: http://proceedings.mlr.press/v97/lange19a/lange19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lange19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Joong-Ho
family: Won
- given: Jason
family: Xu
- given: Kenneth
family: Lange
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3642-3651
id: lange19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3642
lastpage: 3651
published: 2019-05-24 00:00:00 +0000
- title: 'Safe Policy Improvement with Baseline Bootstrapping'
abstract: 'This paper considers Safe Policy Improvement (SPI) in Batch Reinforcement Learning (Batch RL): from a fixed dataset and without direct access to the true environment, train a policy that is guaranteed to perform at least as well as the baseline policy used to collect the data. Our approach, called SPI with Baseline Bootstrapping (SPIBB), is inspired by the knows-what-it-knows paradigm: it bootstraps the trained policy with the baseline when the uncertainty is high. Our first algorithm, $\Pi_b$-SPIBB, comes with SPI theoretical guarantees. We also implement a variant, $\Pi_{\leq b}$-SPIBB, that is even more efficient in practice. We apply our algorithms to a motivational stochastic gridworld domain and further demonstrate on randomly generated MDPs the superiority of SPIBB with respect to existing algorithms, not only in safety but also in mean performance. Finally, we implement a model-free version of SPIBB and show its benefits on a navigation task with deep RL implementation called SPIBB-DQN, which is, to the best of our knowledge, the first RL algorithm relying on a neural network representation able to train efficiently and reliably from batch data, without any interaction with the environment.'
volume: 97
URL: https://proceedings.mlr.press/v97/laroche19a.html
PDF: http://proceedings.mlr.press/v97/laroche19a/laroche19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-laroche19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Romain
family: Laroche
- given: Paul
family: Trichelair
- given: Remi Tachet Des
family: Combes
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3652-3661
id: laroche19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3652
lastpage: 3661
published: 2019-05-24 00:00:00 +0000
- title: 'A Better k-means++ Algorithm via Local Search'
abstract: 'In this paper, we develop a new variant of k-means++ seeding that in expectation achieves a constant approximation guarantee. We obtain this result by a simple combination of k-means++ sampling with a local search strategy. We evaluate our algorithm empirically and show that it also improves the quality of a solution in practice.'
volume: 97
URL: https://proceedings.mlr.press/v97/lattanzi19a.html
PDF: http://proceedings.mlr.press/v97/lattanzi19a/lattanzi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lattanzi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Silvio
family: Lattanzi
- given: Christian
family: Sohler
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3662-3671
id: lattanzi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3662
lastpage: 3671
published: 2019-05-24 00:00:00 +0000
- title: 'Lorentzian Distance Learning for Hyperbolic Representations'
abstract: 'We introduce an approach to learn representations based on the Lorentzian distance in hyperbolic geometry. Hyperbolic geometry is especially suited to hierarchically-structured datasets, which are prevalent in the real world. Current hyperbolic representation learning methods compare examples with the Poincaré distance. They try to minimize the distance of each node in a hierarchy with its descendants while maximizing its distance with other nodes. This formulation produces node representations close to the centroid of their descendants. To obtain efficient and interpretable algorithms, we exploit the fact that the centroid w.r.t the squared Lorentzian distance can be written in closed-form. We show that the Euclidean norm of such a centroid decreases as the curvature of the hyperbolic space decreases. This property makes it appropriate to represent hierarchies where parent nodes minimize the distances to their descendants and have smaller Euclidean norm than their children. Our approach obtains state-of-the-art results in retrieval and classification tasks on different datasets.'
volume: 97
URL: https://proceedings.mlr.press/v97/law19a.html
PDF: http://proceedings.mlr.press/v97/law19a/law19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-law19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Marc
family: Law
- given: Renjie
family: Liao
- given: Jake
family: Snell
- given: Richard
family: Zemel
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3672-3681
id: law19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3672
lastpage: 3681
published: 2019-05-24 00:00:00 +0000
- title: 'DP-GP-LVM: A Bayesian Non-Parametric Model for Learning Multivariate Dependency Structures'
abstract: 'We present a non-parametric Bayesian latent variable model capable of learning dependency structures across dimensions in a multivariate setting. Our approach is based on flexible Gaussian process priors for the generative mappings and interchangeable Dirichlet process priors to learn the structure. The introduction of the Dirichlet process as a specific structural prior allows our model to circumvent issues associated with previous Gaussian process latent variable models. Inference is performed by deriving an efficient variational bound on the marginal log-likelihood of the model. We demonstrate the efficacy of our approach via analysis of discovered structure and superior quantitative performance on missing data imputation.'
volume: 97
URL: https://proceedings.mlr.press/v97/lawrence19a.html
PDF: http://proceedings.mlr.press/v97/lawrence19a/lawrence19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lawrence19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Andrew
family: Lawrence
- given: Carl Henrik
family: Ek
- given: Neill
family: Campbell
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3682-3691
id: lawrence19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3682
lastpage: 3691
published: 2019-05-24 00:00:00 +0000
- title: 'POLITEX: Regret Bounds for Policy Iteration using Expert Prediction'
abstract: 'We present POLITEX (POLicy ITeration with EXpert advice), a variant of policy iteration where each policy is a Boltzmann distribution over the sum of action-value function estimates of the previous policies, and analyze its regret in continuing RL problems. We assume that the value function error after running a policy for $\tau$ time steps scales as $\epsilon(\tau) = \epsilon_0 + O(\sqrt{d/\tau})$, where $\epsilon_0$ is the worst-case approximation error and $d$ is the number of features in a compressed representation of the state-action space. We establish that this condition is satisfied by the LSPE algorithm under certain assumptions on the MDP and policies. Under the error assumption, we show that the regret of POLITEX in uniformly mixing MDPs scales as $O(d^{1/2}T^{3/4} + \epsilon_0T)$, where $O(\cdot)$ hides logarithmic terms and problem-dependent constants. Thus, we provide the first regret bound for a fully practical model-free method which only scales in the number of features, and not in the size of the underlying MDP. Experiments on a queuing problem confirm that POLITEX is competitive with some of its alternatives, while preliminary results on Ms Pacman (one of the standard Atari benchmark problems) confirm the viability of POLITEX beyond linear function approximation.'
volume: 97
URL: https://proceedings.mlr.press/v97/lazic19a.html
PDF: http://proceedings.mlr.press/v97/lazic19a/lazic19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lazic19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yasin
family: Abbasi-Yadkori
- given: Peter
family: Bartlett
- given: Kush
family: Bhatia
- given: Nevena
family: Lazic
- given: Csaba
family: Szepesvari
- given: Gellert
family: Weisz
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3692-3702
id: lazic19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3692
lastpage: 3702
published: 2019-05-24 00:00:00 +0000
- title: 'Batch Policy Learning under Constraints'
abstract: 'When learning policies for real-world domains, two important questions arise: (i) how to efficiently use pre-collected off-policy, non-optimal behavior data; and (ii) how to mediate among different competing objectives and constraints. We thus study the problem of batch policy learning under multiple constraints, and offer a systematic solution. We first propose a flexible meta-algorithm that admits any batch reinforcement learning and online learning procedure as subroutines. We then present a specific algorithmic instantiation and provide performance guarantees for the main objective and all constraints. As part of off-policy learning, we propose a simple method for off-policy policy evaluation (OPE) and derive PAC-style bounds. Our algorithm achieves strong empirical results in different domains, including in a challenging problem of simulated car driving subject to multiple constraints such as lane keeping and smooth driving. We also show experimentally that our OPE method outperforms other popular OPE techniques on a standalone basis, especially in a high-dimensional setting.'
volume: 97
URL: https://proceedings.mlr.press/v97/le19a.html
PDF: http://proceedings.mlr.press/v97/le19a/le19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-le19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Hoang
family: Le
- given: Cameron
family: Voloshin
- given: Yisong
family: Yue
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3703-3712
id: le19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3703
lastpage: 3712
published: 2019-05-24 00:00:00 +0000
- title: 'Target-Based Temporal-Difference Learning'
abstract: 'The use of target networks has been a popular and key component of recent deep Q-learning algorithms for reinforcement learning, yet little is known from the theory side. In this work, we introduce a new family of target-based temporal difference (TD) learning algorithms that maintain two separate learning parameters {–} the target variable and online variable. We propose three members in the family, the averaging TD, double TD, and periodic TD, where the target variable is updated through an averaging, symmetric, or periodic fashion, respectively, mirroring those techniques used in deep Q-learning practice. We establish asymptotic convergence analyses for both averaging TD and double TD and a finite sample analysis for periodic TD. In addition, we provide some simulation results showing potentially superior convergence of these target-based TD algorithms compared to the standard TD-learning. While this work focuses on linear function approximation and policy evaluation setting, we consider this as a meaningful step towards the theoretical understanding of deep Q-learning variants with target networks.'
volume: 97
URL: https://proceedings.mlr.press/v97/lee19a.html
PDF: http://proceedings.mlr.press/v97/lee19a/lee19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lee19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Donghwan
family: Lee
- given: Niao
family: He
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3713-3722
id: lee19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3713
lastpage: 3722
published: 2019-05-24 00:00:00 +0000
- title: 'Functional Transparency for Structured Data: a Game-Theoretic Approach'
abstract: 'We provide a new approach to training neural models to exhibit transparency in a well-defined, functional manner. Our approach naturally operates over structured data and tailors the predictor, functionally, towards a chosen family of (local) witnesses. The estimation problem is setup as a co-operative game between an unrestricted *predictor* such as a neural network, and a set of *witnesses* chosen from the desired transparent family. The goal of the witnesses is to highlight, locally, how well the predictor conforms to the chosen family of functions, while the predictor is trained to minimize the highlighted discrepancy. We emphasize that the predictor remains globally powerful as it is only encouraged to agree locally with locally adapted witnesses. We analyze the effect of the proposed approach, provide example formulations in the context of deep graph and sequence models, and empirically illustrate the idea in chemical property prediction, temporal modeling, and molecule representation learning.'
volume: 97
URL: https://proceedings.mlr.press/v97/lee19b.html
PDF: http://proceedings.mlr.press/v97/lee19b/lee19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lee19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Guang-He
family: Lee
- given: Wengong
family: Jin
- given: David
family: Alvarez-Melis
- given: Tommi
family: Jaakkola
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3723-3733
id: lee19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3723
lastpage: 3733
published: 2019-05-24 00:00:00 +0000
- title: 'Self-Attention Graph Pooling'
abstract: 'Advanced methods of applying deep learning to structured data such as graphs have been proposed in recent years. In particular, studies have focused on generalizing convolutional neural networks to graph data, which includes redefining the convolution and the downsampling (pooling) operations for graphs. The method of generalizing the convolution operation to graphs has been proven to improve performance and is widely used. However, the method of applying downsampling to graphs is still difficult to perform and has room for improvement. In this paper, we propose a graph pooling method based on self-attention. Self-attention using graph convolution allows our pooling method to consider both node features and graph topology. To ensure a fair comparison, the same training procedures and model architectures were used for the existing pooling methods and our method. The experimental results demonstrate that our method achieves superior graph classification performance on the benchmark datasets using a reasonable number of parameters.'
volume: 97
URL: https://proceedings.mlr.press/v97/lee19c.html
PDF: http://proceedings.mlr.press/v97/lee19c/lee19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lee19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Junhyun
family: Lee
- given: Inyeop
family: Lee
- given: Jaewoo
family: Kang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3734-3743
id: lee19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3734
lastpage: 3743
published: 2019-05-24 00:00:00 +0000
- title: 'Set Transformer: A Framework for Attention-based Permutation-Invariant Neural Networks'
abstract: 'Many machine learning tasks such as multiple instance learning, 3D shape recognition, and few-shot image classification are defined on sets of instances. Since solutions to such problems do not depend on the order of elements of the set, models used to address them should be permutation invariant. We present an attention-based neural network module, the Set Transformer, specifically designed to model interactions among elements in the input set. The model consists of an encoder and a decoder, both of which rely on attention mechanisms. In an effort to reduce computational complexity, we introduce an attention scheme inspired by inducing point methods from sparse Gaussian process literature. It reduces the computation time of self-attention from quadratic to linear in the number of elements in the set. We show that our model is theoretically attractive and we evaluate it on a range of tasks, demonstrating the state-of-the-art performance compared to recent methods for set-structured data.'
volume: 97
URL: https://proceedings.mlr.press/v97/lee19d.html
PDF: http://proceedings.mlr.press/v97/lee19d/lee19d.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lee19d.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Juho
family: Lee
- given: Yoonho
family: Lee
- given: Jungtaek
family: Kim
- given: Adam
family: Kosiorek
- given: Seungjin
family: Choi
- given: Yee Whye
family: Teh
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3744-3753
id: lee19d
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3744
lastpage: 3753
published: 2019-05-24 00:00:00 +0000
- title: 'First-Order Algorithms Converge Faster than $O(1/k)$ on Convex Problems'
abstract: 'It is well known that both gradient descent and stochastic coordinate descent achieve a global convergence rate of $O(1/k)$ in the objective value, when applied to a scheme for minimizing a Lipschitz-continuously differentiable, unconstrained convex function. In this work, we improve this rate to $o(1/k)$. We extend the result to proximal gradient and proximal coordinate descent on regularized problems to show similar $o(1/k)$ convergence rates. The result is tight in the sense that a rate of $O(1/k^{1+\epsilon})$ is not generally attainable for any $\epsilon>0$, for any of these methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/lee19e.html
PDF: http://proceedings.mlr.press/v97/lee19e/lee19e.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lee19e.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ching-Pei
family: Lee
- given: Stephen
family: Wright
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3754-3762
id: lee19e
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3754
lastpage: 3762
published: 2019-05-24 00:00:00 +0000
- title: 'Robust Inference via Generative Classifiers for Handling Noisy Labels'
abstract: 'Large-scale datasets may contain significant proportions of noisy (incorrect) class labels, and it is well-known that modern deep neural networks (DNNs) poorly generalize from such noisy training datasets. To mitigate the issue, we propose a novel inference method, termed Robust Generative classifier (RoG), applicable to any discriminative (e.g., softmax) neural classifier pre-trained on noisy datasets. In particular, we induce a generative classifier on top of hidden feature spaces of the pre-trained DNNs, for obtaining a more robust decision boundary. By estimating the parameters of generative classifier using the minimum covariance determinant estimator, we significantly improve the classification accuracy with neither re-training of the deep model nor changing its architectures. With the assumption of Gaussian distribution for features, we prove that RoG generalizes better than baselines under noisy labels. Finally, we propose the ensemble version of RoG to improve its performance by investigating the layer-wise characteristics of DNNs. Our extensive experimental results demonstrate the superiority of RoG given different learning models optimized by several training techniques to handle diverse scenarios of noisy labels.'
volume: 97
URL: https://proceedings.mlr.press/v97/lee19f.html
PDF: http://proceedings.mlr.press/v97/lee19f/lee19f.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lee19f.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Kimin
family: Lee
- given: Sukmin
family: Yun
- given: Kibok
family: Lee
- given: Honglak
family: Lee
- given: Bo
family: Li
- given: Jinwoo
family: Shin
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3763-3772
id: lee19f
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3763
lastpage: 3772
published: 2019-05-24 00:00:00 +0000
- title: 'Sublinear Time Nearest Neighbor Search over Generalized Weighted Space'
abstract: 'Nearest Neighbor Search (NNS) over generalized weighted space is a fundamental problem which has many applications in various fields. However, to the best of our knowledge, there is no sublinear time solution to this problem. Based on the idea of Asymmetric Locality-Sensitive Hashing (ALSH), we introduce a novel spherical asymmetric transformation and propose the first two novel weight-oblivious hashing schemes SL-ALSH and S2-ALSH accordingly. We further show that both schemes enjoy a quality guarantee and can answer the NNS queries in sublinear time. Evaluations over three real datasets demonstrate the superior performance of the two proposed schemes.'
volume: 97
URL: https://proceedings.mlr.press/v97/lei19a.html
PDF: http://proceedings.mlr.press/v97/lei19a/lei19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lei19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yifan
family: Lei
- given: Qiang
family: Huang
- given: Mohan
family: Kankanhalli
- given: Anthony
family: Tung
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3773-3781
id: lei19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3773
lastpage: 3781
published: 2019-05-24 00:00:00 +0000
- title: 'MONK Outlier-Robust Mean Embedding Estimation by Median-of-Means'
abstract: 'Mean embeddings provide an extremely flexible and powerful tool in machine learning and statistics to represent probability distributions and define a semi-metric (MMD, maximum mean discrepancy; also called N-distance or energy distance), with numerous successful applications. The representation is constructed as the expectation of the feature map defined by a kernel. As a mean, its classical empirical estimator, however, can be arbitrary severely affected even by a single outlier in case of unbounded features. To the best of our knowledge, unfortunately even the consistency of the existing few techniques trying to alleviate this serious sensitivity bottleneck is unknown. In this paper, we show how the recently emerged principle of median-of-means can be used to design estimators for kernel mean embedding and MMD with excessive resistance properties to outliers, and optimal sub-Gaussian deviation bounds under mild assumptions.'
volume: 97
URL: https://proceedings.mlr.press/v97/lerasle19a.html
PDF: http://proceedings.mlr.press/v97/lerasle19a/lerasle19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lerasle19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Matthieu
family: Lerasle
- given: Zoltan
family: Szabo
- given: Timothée
family: Mathieu
- given: Guillaume
family: Lecue
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3782-3793
id: lerasle19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3782
lastpage: 3793
published: 2019-05-24 00:00:00 +0000
- title: 'Cheap Orthogonal Constraints in Neural Networks: A Simple Parametrization of the Orthogonal and Unitary Group'
abstract: 'We introduce a novel approach to perform first-order optimization with orthogonal and unitary constraints. This approach is based on a parametrization stemming from Lie group theory through the exponential map. The parametrization transforms the constrained optimization problem into an unconstrained one over a Euclidean space, for which common first-order optimization methods can be used. The theoretical results presented are general enough to cover the special orthogonal group, the unitary group and, in general, any connected compact Lie group. We discuss how this and other parametrizations can be computed efficiently through an implementation trick, making numerically complex parametrizations usable at a negligible runtime cost in neural networks. In particular, we apply our results to RNNs with orthogonal recurrent weights, yielding a new architecture called expRNN. We demonstrate how our method constitutes a more robust approach to optimization with orthogonal constraints, showing faster, accurate, and more stable convergence in several tasks designed to test RNNs.'
volume: 97
URL: https://proceedings.mlr.press/v97/lezcano-casado19a.html
PDF: http://proceedings.mlr.press/v97/lezcano-casado19a/lezcano-casado19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lezcano-casado19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Mario
family: Lezcano-Casado
- given: David
family: Martı́nez-Rubio
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3794-3803
id: lezcano-casado19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3794
lastpage: 3803
published: 2019-05-24 00:00:00 +0000
- title: 'Are Generative Classifiers More Robust to Adversarial Attacks?'
abstract: 'There is a rising interest in studying the robustness of deep neural network classifiers against adversaries, with both advanced attack and defence techniques being actively developed. However, most recent work focuses on discriminative classifiers, which only model the conditional distribution of the labels given the inputs. In this paper, we propose and investigate the deep Bayes classifier, which improves classical naive Bayes with conditional deep generative models. We further develop detection methods for adversarial examples, which reject inputs with low likelihood under the generative model. Experimental results suggest that deep Bayes classifiers are more robust than deep discriminative classifiers, and that the proposed detection methods are effective against many recently proposed attacks.'
volume: 97
URL: https://proceedings.mlr.press/v97/li19a.html
PDF: http://proceedings.mlr.press/v97/li19a/li19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-li19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yingzhen
family: Li
- given: John
family: Bradshaw
- given: Yash
family: Sharma
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3804-3814
id: li19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3804
lastpage: 3814
published: 2019-05-24 00:00:00 +0000
- title: 'Sublinear quantum algorithms for training linear and kernel-based classifiers'
abstract: 'We investigate quantum algorithms for classification, a fundamental problem in machine learning, with provable guarantees. Given $n$ $d$-dimensional data points, the state-of-the-art (and optimal) classical algorithm for training classifiers with constant margin by Clarkson et al. runs in $\tilde{O}(n +d)$, which is also optimal in its input/output model. We design sublinear quantum algorithms for the same task running in $\tilde{O}(\sqrt{n} +\sqrt{d})$, a quadratic improvement in both $n$ and $d$. Moreover, our algorithms use the standard quantization of the classical input and generate the same classical output, suggesting minimal overheads when used as subroutines for end-to-end applications. We also demonstrate a tight lower bound (up to poly-log factors) and discuss the possibility of implementation on near-term quantum machines.'
volume: 97
URL: https://proceedings.mlr.press/v97/li19b.html
PDF: http://proceedings.mlr.press/v97/li19b/li19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-li19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Tongyang
family: Li
- given: Shouvanik
family: Chakrabarti
- given: Xiaodi
family: Wu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3815-3824
id: li19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3815
lastpage: 3824
published: 2019-05-24 00:00:00 +0000
- title: 'LGM-Net: Learning to Generate Matching Networks for Few-Shot Learning'
abstract: 'In this work, we propose a novel meta-learning approach for few-shot classification, which learns transferable prior knowledge across tasks and directly produces network parameters for similar unseen tasks with training samples. Our approach, called LGM-Net, includes two key modules, namely, TargetNet and MetaNet. The TargetNet module is a neural network for solving a specific task and the MetaNet module aims at learning to generate functional weights for TargetNet by observing training samples. We also present an intertask normalization strategy for the training process to leverage common information shared across different tasks. The experimental results on Omniglot and miniImageNet datasets demonstrate that LGM-Net can effectively adapt to similar unseen tasks and achieve competitive performance, and the results on synthetic datasets show that transferable prior knowledge is learned by the MetaNet module via mapping training data to functional weights. LGM-Net enables fast learning and adaptation since no further tuning steps are required compared to other meta-learning approaches'
volume: 97
URL: https://proceedings.mlr.press/v97/li19c.html
PDF: http://proceedings.mlr.press/v97/li19c/li19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-li19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Huaiyu
family: Li
- given: Weiming
family: Dong
- given: Xing
family: Mei
- given: Chongyang
family: Ma
- given: Feiyue
family: Huang
- given: Bao-Gang
family: Hu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3825-3834
id: li19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3825
lastpage: 3834
published: 2019-05-24 00:00:00 +0000
- title: 'Graph Matching Networks for Learning the Similarity of Graph Structured Objects'
abstract: 'This paper addresses the challenging problem of retrieval and matching of graph structured objects, and makes two key contributions. First, we demonstrate how Graph Neural Networks (GNN), which have emerged as an effective model for various supervised prediction problems defined on structured data, can be trained to produce embedding of graphs in vector spaces that enables efficient similarity reasoning. Second, we propose a novel Graph Matching Network model that, given a pair of graphs as input, computes a similarity score between them by jointly reasoning on the pair through a new cross-graph attention-based matching mechanism. We demonstrate the effectiveness of our models on different domains including the challenging problem of control-flow graph based function similarity search that plays an important role in the detection of vulnerabilities in software systems. The experimental analysis demonstrates that our models are not only able to exploit structure in the context of similarity learning but they can also outperform domain specific baseline systems that have been carefully hand-engineered for these problems.'
volume: 97
URL: https://proceedings.mlr.press/v97/li19d.html
PDF: http://proceedings.mlr.press/v97/li19d/li19d.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-li19d.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yujia
family: Li
- given: Chenjie
family: Gu
- given: Thomas
family: Dullien
- given: Oriol
family: Vinyals
- given: Pushmeet
family: Kohli
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3835-3845
id: li19d
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3835
lastpage: 3845
published: 2019-05-24 00:00:00 +0000
- title: 'Area Attention'
abstract: 'Existing attention mechanisms are trained to attend to individual items in a collection (the memory) with a predefined, fixed granularity, e.g., a word token or an image grid. We propose area attention: a way to attend to areas in the memory, where each area contains a group of items that are structurally adjacent, e.g., spatially for a 2D memory such as images, or temporally for a 1D memory such as natural language sentences. Importantly, the shape and the size of an area are dynamically determined via learning, which enables a model to attend to information with varying granularity. Area attention can easily work with existing model architectures such as multi-head attention for simultaneously attending to multiple areas in the memory. We evaluate area attention on two tasks: neural machine translation (both character and token-level) and image captioning, and improve upon strong (state-of-the-art) baselines in all the cases. These improvements are obtainable with a basic form of area attention that is parameter free.'
volume: 97
URL: https://proceedings.mlr.press/v97/li19e.html
PDF: http://proceedings.mlr.press/v97/li19e/li19e.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-li19e.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yang
family: Li
- given: Lukasz
family: Kaiser
- given: Samy
family: Bengio
- given: Si
family: Si
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3846-3855
id: li19e
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3846
lastpage: 3855
published: 2019-05-24 00:00:00 +0000
- title: 'Online Learning to Rank with Features'
abstract: 'We introduce a new model for online ranking in which the click probability factors into an examination and attractiveness function and the attractiveness function is a linear function of a feature vector and an unknown parameter. Only relatively mild assumptions are made on the examination function. A novel algorithm for this setup is analysed, showing that the dependence on the number of items is replaced by a dependence on the dimension, allowing the new algorithm to handle a large number of items. When reduced to the orthogonal case, the regret of the algorithm improves on the state-of-the-art.'
volume: 97
URL: https://proceedings.mlr.press/v97/li19f.html
PDF: http://proceedings.mlr.press/v97/li19f/li19f.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-li19f.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Shuai
family: Li
- given: Tor
family: Lattimore
- given: Csaba
family: Szepesvari
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3856-3865
id: li19f
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3856
lastpage: 3865
published: 2019-05-24 00:00:00 +0000
- title: 'NATTACK: Learning the Distributions of Adversarial Examples for an Improved Black-Box Attack on Deep Neural Networks'
abstract: 'Powerful adversarial attack methods are vital for understanding how to construct robust deep neural networks (DNNs) and for thoroughly testing defense techniques. In this paper, we propose a black-box adversarial attack algorithm that can defeat both vanilla DNNs and those generated by various defense techniques developed recently. Instead of searching for an "optimal" adversarial example for a benign input to a targeted DNN, our algorithm finds a probability density distribution over a small region centered around the input, such that a sample drawn from this distribution is likely an adversarial example, without the need of accessing the DNN’s internal layers or weights. Our approach is universal as it can successfully attack different neural networks by a single algorithm. It is also strong; according to the testing against 2 vanilla DNNs and 13 defended ones, it outperforms state-of-the-art black-box or white-box attack methods for most test cases. Additionally, our results reveal that adversarial training remains one of the best defense techniques, and the adversarial examples are not as transferable across defended DNNs as them across vanilla DNNs.'
volume: 97
URL: https://proceedings.mlr.press/v97/li19g.html
PDF: http://proceedings.mlr.press/v97/li19g/li19g.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-li19g.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yandong
family: Li
- given: Lijun
family: Li
- given: Liqiang
family: Wang
- given: Tong
family: Zhang
- given: Boqing
family: Gong
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3866-3876
id: li19g
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3866
lastpage: 3876
published: 2019-05-24 00:00:00 +0000
- title: 'Bayesian Joint Spike-and-Slab Graphical Lasso'
abstract: 'In this article, we propose a new class of priors for Bayesian inference with multiple Gaussian graphical models. We introduce Bayesian treatments of two popular procedures, the group graphical lasso and the fused graphical lasso, and extend them to a continuous spike-and-slab framework to allow self-adaptive shrinkage and model selection simultaneously. We develop an EM algorithm that performs fast and dynamic explorations of posterior modes. Our approach selects sparse models efficiently and automatically with substantially smaller bias than would be induced by alternative regularization procedures. The performance of the proposed methods are demonstrated through simulation and two real data examples.'
volume: 97
URL: https://proceedings.mlr.press/v97/li19h.html
PDF: http://proceedings.mlr.press/v97/li19h/li19h.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-li19h.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Zehang
family: Li
- given: Tyler
family: Mccormick
- given: Samuel
family: Clark
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3877-3885
id: li19h
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3877
lastpage: 3885
published: 2019-05-24 00:00:00 +0000
- title: 'Exploiting Worker Correlation for Label Aggregation in Crowdsourcing'
abstract: 'Crowdsourcing has emerged as a core component of data science pipelines. From collected noisy worker labels, aggregation models that incorporate worker reliability parameters aim to infer a latent true annotation. In this paper, we argue that existing crowdsourcing approaches do not sufficiently model worker correlations observed in practical settings; we propose in response an enhanced Bayesian classifier combination (EBCC) model, with inference based on a mean-field variational approach. An introduced mixture of intra-class reliabilities—connected to tensor decomposition and item clustering—induces inter-worker correlation. EBCC does not suffer the limitations of existing correlation models: intractable marginalisation of missing labels and poor scaling to large worker cohorts. Extensive empirical comparison on 17 real-world datasets sees EBCC achieving the highest mean accuracy across 10 benchmark crowdsourcing methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/li19i.html
PDF: http://proceedings.mlr.press/v97/li19i/li19i.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-li19i.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yuan
family: Li
- given: Benjamin
family: Rubinstein
- given: Trevor
family: Cohn
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3886-3895
id: li19i
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3886
lastpage: 3895
published: 2019-05-24 00:00:00 +0000
- title: 'Adversarial camera stickers: A physical camera-based attack on deep learning systems'
abstract: 'Recent work has documented the susceptibility of deep learning systems to adversarial examples, but most such attacks directly manipulate the digital input to a classifier. Although a smaller line of work considers physical adversarial attacks, in all cases these involve manipulating the object of interest, e.g., putting a physical sticker on an object to misclassify it, or manufacturing an object specifically intended to be misclassified. In this work, we consider an alternative question: is it possible to fool deep classifiers, over all perceived objects of a certain type, by physically manipulating the camera itself? We show that by placing a carefully crafted and mainly-translucent sticker over the lens of a camera, one can create universal perturbations of the observed images that are inconspicuous, yet misclassify target objects as a different (targeted) class. To accomplish this, we propose an iterative procedure for both updating the attack perturbation (to make it adversarial for a given classifier), and the threat model itself (to ensure it is physically realizable). For example, we show that we can achieve physically-realizable attacks that fool ImageNet classifiers in a targeted fashion 49.6% of the time. This presents a new class of physically-realizable threat models to consider in the context of adversarially robust machine learning. Our demo video can be viewed at: https://youtu.be/wUVmL33Fx54'
volume: 97
URL: https://proceedings.mlr.press/v97/li19j.html
PDF: http://proceedings.mlr.press/v97/li19j/li19j.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-li19j.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Juncheng
family: Li
- given: Frank
family: Schmidt
- given: Zico
family: Kolter
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3896-3904
id: li19j
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3896
lastpage: 3904
published: 2019-05-24 00:00:00 +0000
- title: 'Towards a Unified Analysis of Random Fourier Features'
abstract: 'Random Fourier features is a widely used, simple, and effective technique for scaling up kernel methods. The existing theoretical analysis of the approach, however, remains focused on specific learning tasks and typically gives pessimistic bounds which are at odds with the empirical results. We tackle these problems and provide the first unified risk analysis of learning with random Fourier features using the squared error and Lipschitz continuous loss functions. In our bounds, the trade-off between the computational cost and the expected risk convergence rate is problem specific and expressed in terms of the regularization parameter and the *number of effective degrees of freedom*. We study both the standard random Fourier features method for which we improve the existing bounds on the number of features required to guarantee the corresponding minimax risk convergence rate of kernel ridge regression, as well as a data-dependent modification which samples features proportional to *ridge leverage scores* and further reduces the required number of features. As ridge leverage scores are expensive to compute, we devise a simple approximation scheme which provably reduces the computational cost without loss of statistical efficiency.'
volume: 97
URL: https://proceedings.mlr.press/v97/li19k.html
PDF: http://proceedings.mlr.press/v97/li19k/li19k.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-li19k.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Zhu
family: Li
- given: Jean-Francois
family: Ton
- given: Dino
family: Oglic
- given: Dino
family: Sejdinovic
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3905-3914
id: li19k
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3905
lastpage: 3914
published: 2019-05-24 00:00:00 +0000
- title: 'Feature-Critic Networks for Heterogeneous Domain Generalization'
abstract: 'The well known domain shift issue causes model performance to degrade when deployed to a new target domain with different statistics to training. Domain adaptation techniques alleviate this, but need some instances from the target domain to drive adaptation. Domain generalisation is the recently topical problem of learning a model that generalises to unseen domains out of the box, and various approaches aim to train a domain-invariant feature extractor, typically by adding some manually designed losses. In this work, we propose a learning to learn approach, where the auxiliary loss that helps generalisation is itself learned. Beyond conventional domain generalisation, we consider a more challenging setting of heterogeneous domain generalisation, where the unseen domains do not share label space with the seen ones, and the goal is to train a feature representation that is useful off-the-shelf for novel data and novel categories. Experimental evaluation demonstrates that our method outperforms state-of-the-art solutions in both settings.'
volume: 97
URL: https://proceedings.mlr.press/v97/li19l.html
PDF: http://proceedings.mlr.press/v97/li19l/li19l.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-li19l.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yiying
family: Li
- given: Yongxin
family: Yang
- given: Wei
family: Zhou
- given: Timothy
family: Hospedales
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3915-3924
id: li19l
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3915
lastpage: 3924
published: 2019-05-24 00:00:00 +0000
- title: 'Learn to Grow: A Continual Structure Learning Framework for Overcoming Catastrophic Forgetting'
abstract: 'Addressing catastrophic forgetting is one of the key challenges in continual learning where machine learning systems are trained with sequential or streaming tasks. Despite recent remarkable progress in state-of-the-art deep learning, deep neural networks (DNNs) are still plagued with the catastrophic forgetting problem. This paper presents a conceptually simple yet general and effective framework for handling catastrophic forgetting in continual learning with DNNs. The proposed method consists of two components: a neural structure optimization component and a parameter learning and/or fine-tuning component. By separating the explicit neural structure learning and the parameter estimation, not only is the proposed method capable of evolving neural structures in an intuitively meaningful way, but also shows strong capabilities of alleviating catastrophic forgetting in experiments. Furthermore, the proposed method outperforms all other baselines on the permuted MNIST dataset, the split CIFAR100 dataset and the Visual Domain Decathlon dataset in continual learning setting.'
volume: 97
URL: https://proceedings.mlr.press/v97/li19m.html
PDF: http://proceedings.mlr.press/v97/li19m/li19m.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-li19m.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Xilai
family: Li
- given: Yingbo
family: Zhou
- given: Tianfu
family: Wu
- given: Richard
family: Socher
- given: Caiming
family: Xiong
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3925-3934
id: li19m
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3925
lastpage: 3934
published: 2019-05-24 00:00:00 +0000
- title: 'Alternating Minimizations Converge to Second-Order Optimal Solutions'
abstract: 'This work studies the second-order convergence for both standard alternating minimization and proximal alternating minimization. We show that under mild assumptions on the (nonconvex) objective function, both algorithms avoid strict saddles almost surely from random initialization. Together with known first-order convergence results, this implies both algorithms converge to a second-order stationary point. This solves an open problem for the second-order convergence of alternating minimization algorithms that have been widely used in practice to solve large-scale nonconvex problems due to their simple implementation, fast convergence, and superb empirical performance.'
volume: 97
URL: https://proceedings.mlr.press/v97/li19n.html
PDF: http://proceedings.mlr.press/v97/li19n/li19n.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-li19n.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Qiuwei
family: Li
- given: Zhihui
family: Zhu
- given: Gongguo
family: Tang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3935-3943
id: li19n
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3935
lastpage: 3943
published: 2019-05-24 00:00:00 +0000
- title: 'Cautious Regret Minimization: Online Optimization with Long-Term Budget Constraints'
abstract: 'We study a class of online convex optimization problems with long-term budget constraints that arise naturally as reliability guarantees or total consumption constraints. In this general setting, prior work by Mannor et al. (2009) has shown that achieving no regret is impossible if the functions defining the agent’s budget are chosen by an adversary. To overcome this obstacle, we refine the agent’s regret metric by introducing the notion of a "K-benchmark", i.e., a comparator which meets the problem’s allotted budget over any window of length K. The impossibility analysis of Mannor et al. (2009) is recovered when K=T; however, for K=o(T), we show that it is possible to minimize regret while still meeting the problem’s long-term budget constraints. We achieve this via an online learning policy based on Cautious Online Lagrangiant Descent (COLD) for which we derive explicit bounds, in terms of both the incurred regret and the residual budget violations.'
volume: 97
URL: https://proceedings.mlr.press/v97/liakopoulos19a.html
PDF: http://proceedings.mlr.press/v97/liakopoulos19a/liakopoulos19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-liakopoulos19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Nikolaos
family: Liakopoulos
- given: Apostolos
family: Destounis
- given: Georgios
family: Paschos
- given: Thrasyvoulos
family: Spyropoulos
- given: Panayotis
family: Mertikopoulos
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3944-3952
id: liakopoulos19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3944
lastpage: 3952
published: 2019-05-24 00:00:00 +0000
- title: 'Regularization in directable environments with application to Tetris'
abstract: 'Learning from small data sets is difficult in the absence of specific domain knowledge. We present a regularized linear model called STEW that benefits from a generic and prevalent form of prior knowledge: feature directions. STEW shrinks weights toward each other, converging to an equal-weights solution in the limit of infinite regularization. We provide theoretical results on the equal-weights solution that explains how STEW can productively trade-off bias and variance. Across a wide range of learning problems, including Tetris, STEW outperformed existing linear models, including ridge regression, the Lasso, and the non-negative Lasso, when feature directions were known. The model proved to be robust to unreliable (or absent) feature directions, still outperforming alternative models under diverse conditions. Our results in Tetris were obtained by using a novel approach to learning in sequential decision environments based on multinomial logistic regression.'
volume: 97
URL: https://proceedings.mlr.press/v97/lichtenberg19a.html
PDF: http://proceedings.mlr.press/v97/lichtenberg19a/lichtenberg19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lichtenberg19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jan Malte
family: Lichtenberg
- given: Özgür
family: Şimşek
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3953-3962
id: lichtenberg19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3953
lastpage: 3962
published: 2019-05-24 00:00:00 +0000
- title: 'Inference and Sampling of $K_33$-free Ising Models'
abstract: 'We call an Ising model tractable when it is possible to compute its partition function value (statistical inference) in polynomial time. The tractability also implies an ability to sample configurations of this model in polynomial time. The notion of tractability extends the basic case of planar zero-field Ising models. Our starting point is to describe algorithms for the basic case, computing partition function and sampling efficiently. Then, we extend our tractable inference and sampling algorithms to models whose triconnected components are either planar or graphs of $O(1)$ size. In particular, it results in a polynomial-time inference and sampling algorithms for $K_{33}$ (minor)-free topologies of zero-field Ising models—a generalization of planar graphs with a potentially unbounded genus.'
volume: 97
URL: https://proceedings.mlr.press/v97/likhosherstov19a.html
PDF: http://proceedings.mlr.press/v97/likhosherstov19a/likhosherstov19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-likhosherstov19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Valerii
family: Likhosherstov
- given: Yury
family: Maximov
- given: Misha
family: Chertkov
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3963-3972
id: likhosherstov19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3963
lastpage: 3972
published: 2019-05-24 00:00:00 +0000
- title: 'Kernel-Based Reinforcement Learning in Robust Markov Decision Processes'
abstract: 'The robust Markov decision processes (MDP) framework aims to address the problem of parameter uncertainty due to model mismatch, approximation errors or even adversarial behaviors. It is especially relevant when deploying the learned policies in real-world applications. Scaling up the robust MDP framework to large or continuous state space remains a challenging problem. The use of function approximation in this case is usually inevitable and this can only amplify the problem of model mismatch and parameter uncertainties. It has been previously shown that, in the case of MDPs with state aggregation, the robust policies enjoy a tighter performance bound compared to standard solutions due to its reduced sensitivity to approximation errors. We extend these results to the much larger class of kernel-based approximators and show, both analytically and empirically that the robust policies can significantly outperform the non-robust counterpart.'
volume: 97
URL: https://proceedings.mlr.press/v97/lim19a.html
PDF: http://proceedings.mlr.press/v97/lim19a/lim19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lim19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Shiau Hong
family: Lim
- given: Arnaud
family: Autef
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3973-3981
id: lim19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3973
lastpage: 3981
published: 2019-05-24 00:00:00 +0000
- title: 'On Efficient Optimal Transport: An Analysis of Greedy and Accelerated Mirror Descent Algorithms'
abstract: 'We provide theoretical analyses for two algorithms that solve the regularized optimal transport (OT) problem between two discrete probability measures with at most $n$ atoms. We show that a greedy variant of the classical Sinkhorn algorithm, known as the *Greenkhorn algorithm*, can be improved to $\bigOtil\left(n^2/\varepsilon^2\right)$, improving on the best known complexity bound of $\bigOtil\left(n^2/\varepsilon^3\right)$. This matches the best known complexity bound for the Sinkhorn algorithm and helps explain why the Greenkhorn algorithm outperforms the Sinkhorn algorithm in practice. Our proof technique is based on a primal-dual formulation and provide a *tight* upper bound for the dual solution, leading to a class of *adaptive primal-dual accelerated mirror descent* (APDAMD) algorithms. We prove that the complexity of these algorithms is $\bigOtil\left(n^2\sqrt{\gamma}/\varepsilon\right)$ in which $\gamma \in (0, n]$ refers to some constants in the Bregman divergence. Experimental results on synthetic and real datasets demonstrate the favorable performance of the Greenkhorn and APDAMD algorithms in practice.'
volume: 97
URL: https://proceedings.mlr.press/v97/lin19a.html
PDF: http://proceedings.mlr.press/v97/lin19a/lin19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lin19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Tianyi
family: Lin
- given: Nhat
family: Ho
- given: Michael
family: Jordan
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3982-3991
id: lin19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3982
lastpage: 3991
published: 2019-05-24 00:00:00 +0000
- title: 'Fast and Simple Natural-Gradient Variational Inference with Mixture of Exponential-family Approximations'
abstract: 'Natural-gradient methods enable fast and simple algorithms for variational inference, but due to computational difficulties, their use is mostly limited to minimal exponential-family (EF) approximations. In this paper, we extend their application to estimate structured approximations such as mixtures of EF distributions. Such approximations can fit complex, multimodal posterior distributions and are generally more accurate than unimodal EF approximations. By using a minimal conditional-EF representation of such approximations, we derive simple natural-gradient updates. Our empirical results demonstrate a faster convergence of our natural-gradient method compared to black-box gradient-based methods. Our work expands the scope of natural gradients for Bayesian inference and makes them more widely applicable than before.'
volume: 97
URL: https://proceedings.mlr.press/v97/lin19b.html
PDF: http://proceedings.mlr.press/v97/lin19b/lin19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lin19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Wu
family: Lin
- given: Mohammad Emtiyaz
family: Khan
- given: Mark
family: Schmidt
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 3992-4002
id: lin19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 3992
lastpage: 4002
published: 2019-05-24 00:00:00 +0000
- title: 'Acceleration of SVRG and Katyusha X by Inexact Preconditioning'
abstract: 'Empirical risk minimization is an important class of optimization problems with many popular machine learning applications, and stochastic variance reduction methods are popular choices for solving them. Among these methods, SVRG and Katyusha X (a Nesterov accelerated SVRG) achieve fast convergence without substantial memory requirement. In this paper, we propose to accelerate these two algorithms by *inexact preconditioning*, the proposed methods employ *fixed* preconditioners, although the subproblem in each epoch becomes harder, it suffices to apply *fixed* number of simple subroutines to solve it inexactly, without losing the overall convergence. As a result, this inexact preconditioning strategy gives provably better iteration complexity and gradient complexity over SVRG and Katyusha X. We also allow each function in the finite sum to be nonconvex while the sum is strongly convex. In our numerical experiments, we observe an on average $8\times$ speedup on the number of iterations and $7\times$ speedup on runtime.'
volume: 97
URL: https://proceedings.mlr.press/v97/liu19a.html
PDF: http://proceedings.mlr.press/v97/liu19a/liu19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-liu19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yanli
family: Liu
- given: Fei
family: Feng
- given: Wotao
family: Yin
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4003-4012
id: liu19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4003
lastpage: 4012
published: 2019-05-24 00:00:00 +0000
- title: 'Transferable Adversarial Training: A General Approach to Adapting Deep Classifiers'
abstract: 'Domain adaptation enables knowledge transfer from a labeled source domain to an unlabeled target domain. A mainstream approach is adversarial feature adaptation, which learns domain-invariant representations through aligning the feature distributions of both domains. However, a theoretical prerequisite of domain adaptation is the adaptability measured by the expected risk of an ideal joint hypothesis over the source and target domains. In this respect, adversarial feature adaptation may potentially deteriorate the adaptability, since it distorts the original feature distributions when suppressing domain-specific variations. To this end, we propose Transferable Adversarial Training (TAT) to enable the adaptation of deep classifiers. The approach generates transferable examples to fill in the gap between the source and target domains, and adversarially trains the deep classifiers to make consistent predictions over the transferable examples. Without learning domain-invariant representations at the expense of distorting the feature distributions, the adaptability in the theoretical learning bound is algorithmically guaranteed. A series of experiments validate that our approach advances the state of the arts on a variety of domain adaptation tasks in vision and NLP, including object recognition, learning from synthetic to real data, and sentiment classification.'
volume: 97
URL: https://proceedings.mlr.press/v97/liu19b.html
PDF: http://proceedings.mlr.press/v97/liu19b/liu19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-liu19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Hong
family: Liu
- given: Mingsheng
family: Long
- given: Jianmin
family: Wang
- given: Michael
family: Jordan
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4013-4022
id: liu19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4013
lastpage: 4022
published: 2019-05-24 00:00:00 +0000
- title: 'Rao-Blackwellized Stochastic Gradients for Discrete Distributions'
abstract: 'We wish to compute the gradient of an expectation over a finite or countably infinite sample space having K $\leq$ $\infty$ categories. When K is indeed infinite, or finite but very large, the relevant summation is intractable. Accordingly, various stochastic gradient estimators have been proposed. In this paper, we describe a technique that can be applied to reduce the variance of any such estimator, without changing its bias{—}in particular, unbiasedness is retained. We show that our technique is an instance of Rao-Blackwellization, and we demonstrate the improvement it yields on a semi-supervised classification problem and a pixel attention task.'
volume: 97
URL: https://proceedings.mlr.press/v97/liu19c.html
PDF: http://proceedings.mlr.press/v97/liu19c/liu19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-liu19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Runjing
family: Liu
- given: Jeffrey
family: Regier
- given: Nilesh
family: Tripuraneni
- given: Michael
family: Jordan
- given: Jon
family: Mcauliffe
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4023-4031
id: liu19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4023
lastpage: 4031
published: 2019-05-24 00:00:00 +0000
- title: 'Sparse Extreme Multi-label Learning with Oracle Property'
abstract: 'The pioneering work of sparse local embeddings for extreme classification (SLEEC) (Bhatia et al., 2015) has shown great promise in multi-label learning. Unfortunately, the statistical rate of convergence and oracle property of SLEEC are still not well understood. To fill this gap, we present a unified framework for SLEEC with nonconvex penalty. Theoretically, we rigorously prove that our proposed estimator enjoys oracle property (i.e., performs as well as if the underlying model were known beforehand), and obtains a desirable statistical convergence rate. Moreover, we show that under a mild condition on the magnitude of the entries in the underlying model, we are able to obtain an improved convergence rate. Extensive numerical experiments verify our theoretical findings and the superiority of our proposed estimator.'
volume: 97
URL: https://proceedings.mlr.press/v97/liu19d.html
PDF: http://proceedings.mlr.press/v97/liu19d/liu19d.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-liu19d.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Weiwei
family: Liu
- given: Xiaobo
family: Shen
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4032-4041
id: liu19d
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4032
lastpage: 4041
published: 2019-05-24 00:00:00 +0000
- title: 'Data Poisoning Attacks on Stochastic Bandits'
abstract: 'Stochastic multi-armed bandits form a class of online learning problems that have important applications in online recommendation systems, adaptive medical treatment, and many others. Even though potential attacks against these learning algorithms may hijack their behavior, causing catastrophic loss in real-world applications, little is known about adversarial attacks on bandit algorithms. In this paper, we propose a framework of offline attacks on bandit algorithms and study convex optimization based attacks on several popular bandit algorithms. We show that the attacker can force the bandit algorithm to pull a target arm with high probability by a slight manipulation of the rewards in the data. Then we study a form of online attacks on bandit algorithms and propose an adaptive attack strategy against any bandit algorithm without the knowledge of the bandit algorithm. Our adaptive attack strategy can hijack the behavior of the bandit algorithm to suffer a linear regret with only a logarithmic cost to the attacker. Our results demonstrate a significant security threat to stochastic bandits.'
volume: 97
URL: https://proceedings.mlr.press/v97/liu19e.html
PDF: http://proceedings.mlr.press/v97/liu19e/liu19e.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-liu19e.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Fang
family: Liu
- given: Ness
family: Shroff
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4042-4050
id: liu19e
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4042
lastpage: 4050
published: 2019-05-24 00:00:00 +0000
- title: 'The Implicit Fairness Criterion of Unconstrained Learning'
abstract: 'We clarify what fairness guarantees we can and cannot expect to follow from unconstrained machine learning. Specifically, we show that in many settings, unconstrained learning on its own implies group calibration, that is, the outcome variable is conditionally independent of group membership given the score. A lower bound confirms the optimality of our upper bound. Moreover, we prove that as the excess risk of the learned score decreases, the more strongly it violates separation and independence, two other standard fairness criteria. Our results challenge the view that group calibration necessitates an active intervention, suggesting that often we ought to think of it as a byproduct of unconstrained machine learning.'
volume: 97
URL: https://proceedings.mlr.press/v97/liu19f.html
PDF: http://proceedings.mlr.press/v97/liu19f/liu19f.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-liu19f.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Lydia T.
family: Liu
- given: Max
family: Simchowitz
- given: Moritz
family: Hardt
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4051-4060
id: liu19f
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4051
lastpage: 4060
published: 2019-05-24 00:00:00 +0000
- title: 'Taming MAML: Efficient unbiased meta-reinforcement learning'
abstract: 'While meta reinforcement learning (Meta-RL) methods have achieved remarkable success, obtaining correct and low variance estimates for policy gradients remains a significant challenge. In particular, estimating a large Hessian, poor sample efficiency and unstable training continue to make Meta-RL difficult. We propose a surrogate objective function named, Taming MAML (TMAML), that adds control variates into gradient estimation via automatic differentiation. TMAML improves the quality of gradient estimation by reducing variance without introducing bias. We further propose a version of our method that extends the meta-learning framework to learning the control variates themselves, enabling efficient and scalable learning from a distribution of MDPs. We empirically compare our approach with MAML and other variance-bias trade-off methods including DICE, LVC, and action-dependent control variates. Our approach is easy to implement and outperforms existing methods in terms of the variance and accuracy of gradient estimation, ultimately yielding higher performance across a variety of challenging Meta-RL environments.'
volume: 97
URL: https://proceedings.mlr.press/v97/liu19g.html
PDF: http://proceedings.mlr.press/v97/liu19g/liu19g.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-liu19g.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Hao
family: Liu
- given: Richard
family: Socher
- given: Caiming
family: Xiong
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4061-4071
id: liu19g
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4061
lastpage: 4071
published: 2019-05-24 00:00:00 +0000
- title: 'On Certifying Non-Uniform Bounds against Adversarial Attacks'
abstract: 'This work studies the robustness certification problem of neural network models, which aims to find certified adversary-free regions as large as possible around data points. In contrast to the existing approaches that seek regions bounded uniformly along all input features, we consider non-uniform bounds and use it to study the decision boundary of neural network models. We formulate our target as an optimization problem with nonlinear constraints. Then, a framework applicable for general feedforward neural networks is proposed to bound the output logits so that the relaxed problem can be solved by the augmented Lagrangian method. Our experiments show the non-uniform bounds have larger volumes than uniform ones. Compared with normal models, the robust models have even larger non-uniform bounds and better interpretability. Further, the geometric similarity of the non-uniform bounds gives a quantitative, data-agnostic metric of input features’ robustness.'
volume: 97
URL: https://proceedings.mlr.press/v97/liu19h.html
PDF: http://proceedings.mlr.press/v97/liu19h/liu19h.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-liu19h.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Chen
family: Liu
- given: Ryota
family: Tomioka
- given: Volkan
family: Cevher
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4072-4081
id: liu19h
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4072
lastpage: 4081
published: 2019-05-24 00:00:00 +0000
- title: 'Understanding and Accelerating Particle-Based Variational Inference'
abstract: 'Particle-based variational inference methods (ParVIs) have gained attention in the Bayesian inference literature, for their capacity to yield flexible and accurate approximations. We explore ParVIs from the perspective of Wasserstein gradient flows, and make both theoretical and practical contributions. We unify various finite-particle approximations that existing ParVIs use, and recognize that the approximation is essentially a compulsory smoothing treatment, in either of two equivalent forms. This novel understanding reveals the assumptions and relations of existing ParVIs, and also inspires new ParVIs. We propose an acceleration framework and a principled bandwidth-selection method for general ParVIs; these are based on the developed theory and leverage the geometry of the Wasserstein space. Experimental results show the improved convergence by the acceleration framework and enhanced sample accuracy by the bandwidth-selection method.'
volume: 97
URL: https://proceedings.mlr.press/v97/liu19i.html
PDF: http://proceedings.mlr.press/v97/liu19i/liu19i.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-liu19i.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Chang
family: Liu
- given: Jingwei
family: Zhuo
- given: Pengyu
family: Cheng
- given: Ruiyi
family: Zhang
- given: Jun
family: Zhu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4082-4092
id: liu19i
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4082
lastpage: 4092
published: 2019-05-24 00:00:00 +0000
- title: 'Understanding MCMC Dynamics as Flows on the Wasserstein Space'
abstract: 'It is known that the Langevin dynamics used in MCMC is the gradient flow of the KL divergence on the Wasserstein space, which helps convergence analysis and inspires recent particle-based variational inference methods (ParVIs). But no more MCMC dynamics is understood in this way. In this work, by developing novel concepts, we propose a theoretical framework that recognizes a general MCMC dynamics as the fiber-gradient Hamiltonian flow on the Wasserstein space of a fiber-Riemannian Poisson manifold. The "conservation + convergence" structure of the flow gives a clear picture on the behavior of general MCMC dynamics. The framework also enables ParVI simulation of MCMC dynamics, which enriches the ParVI family with more efficient dynamics, and also adapts ParVI advantages to MCMCs. We develop two ParVI methods for a particular MCMC dynamics and demonstrate the benefits in experiments.'
volume: 97
URL: https://proceedings.mlr.press/v97/liu19j.html
PDF: http://proceedings.mlr.press/v97/liu19j/liu19j.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-liu19j.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Chang
family: Liu
- given: Jingwei
family: Zhuo
- given: Jun
family: Zhu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4093-4103
id: liu19j
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4093
lastpage: 4103
published: 2019-05-24 00:00:00 +0000
- title: 'Sliced-Wasserstein Flows: Nonparametric Generative Modeling via Optimal Transport and Diffusions'
abstract: 'By building upon the recent theory that established the connection between implicit generative modeling (IGM) and optimal transport, in this study, we propose a novel parameter-free algorithm for learning the underlying distributions of complicated datasets and sampling from them. The proposed algorithm is based on a functional optimization problem, which aims at finding a measure that is close to the data distribution as much as possible and also expressive enough for generative modeling purposes. We formulate the problem as a gradient flow in the space of probability measures. The connections between gradient flows and stochastic differential equations let us develop a computationally efficient algorithm for solving the optimization problem. We provide formal theoretical analysis where we prove finite-time error guarantees for the proposed algorithm. To the best of our knowledge, the proposed algorithm is the first nonparametric IGM algorithm with explicit theoretical guarantees. Our experimental results support our theory and show that our algorithm is able to successfully capture the structure of different types of data distributions.'
volume: 97
URL: https://proceedings.mlr.press/v97/liutkus19a.html
PDF: http://proceedings.mlr.press/v97/liutkus19a/liutkus19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-liutkus19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Antoine
family: Liutkus
- given: Umut
family: Simsekli
- given: Szymon
family: Majewski
- given: Alain
family: Durmus
- given: Fabian-Robert
family: Stöter
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4104-4113
id: liutkus19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4104
lastpage: 4113
published: 2019-05-24 00:00:00 +0000
- title: 'Challenging Common Assumptions in the Unsupervised Learning of Disentangled Representations'
abstract: 'The key idea behind the unsupervised learning of disentangled representations is that real-world data is generated by a few explanatory factors of variation which can be recovered by unsupervised learning algorithms. In this paper, we provide a sober look at recent progress in the field and challenge some common assumptions. We first theoretically show that the unsupervised learning of disentangled representations is fundamentally impossible without inductive biases on both the models and the data. Then, we train more than $12000$ models covering most prominent methods and evaluation metrics in a reproducible large-scale experimental study on seven different data sets. We observe that while the different methods successfully enforce properties “encouraged” by the corresponding losses, well-disentangled models seemingly cannot be identified without supervision. Furthermore, increased disentanglement does not seem to lead to a decreased sample complexity of learning for downstream tasks. Our results suggest that future work on disentanglement learning should be explicit about the role of inductive biases and (implicit) supervision, investigate concrete benefits of enforcing disentanglement of the learned representations, and consider a reproducible experimental setup covering several data sets.'
volume: 97
URL: https://proceedings.mlr.press/v97/locatello19a.html
PDF: http://proceedings.mlr.press/v97/locatello19a/locatello19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-locatello19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Francesco
family: Locatello
- given: Stefan
family: Bauer
- given: Mario
family: Lucic
- given: Gunnar
family: Raetsch
- given: Sylvain
family: Gelly
- given: Bernhard
family: Schölkopf
- given: Olivier
family: Bachem
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4114-4124
id: locatello19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4114
lastpage: 4124
published: 2019-05-24 00:00:00 +0000
- title: 'Bayesian Counterfactual Risk Minimization'
abstract: 'We present a Bayesian view of counterfactual risk minimization (CRM) for offline learning from logged bandit feedback. Using PAC-Bayesian analysis, we derive a new generalization bound for the truncated inverse propensity score estimator. We apply the bound to a class of Bayesian policies, which motivates a novel, potentially data-dependent, regularization technique for CRM. Experimental results indicate that this technique outperforms standard $L_2$ regularization, and that it is competitive with variance regularization while being both simpler to implement and more computationally efficient.'
volume: 97
URL: https://proceedings.mlr.press/v97/london19a.html
PDF: http://proceedings.mlr.press/v97/london19a/london19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-london19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ben
family: London
- given: Ted
family: Sandler
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4125-4133
id: london19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4125
lastpage: 4133
published: 2019-05-24 00:00:00 +0000
- title: 'PA-GD: On the Convergence of Perturbed Alternating Gradient Descent to Second-Order Stationary Points for Structured Nonconvex Optimization'
abstract: 'Alternating gradient descent (A-GD) is a simple but popular algorithm in machine learning, which updates two blocks of variables in an alternating manner using gradient descent steps. In this paper, we consider a smooth unconstrained nonconvex optimization problem, and propose a perturbed A-GD (PA-GD) which is able to converge (with high probability) to the second-order stationary points (SOSPs) with a global sublinear rate. Existing analysis on A-GD type algorithm either only guarantees convergence to first-order solutions, or converges to second-order solutions asymptotically (without rates). To the best of our knowledge, this is the first alternating type algorithm that takes $\mathcal{O}(\text{polylog}(d)/\epsilon^2)$ iterations to achieve an ($\epsilon,\sqrt{\epsilon}$)-SOSP with high probability, where polylog$(d)$ denotes the polynomial of the logarithm with respect to problem dimension $d$.'
volume: 97
URL: https://proceedings.mlr.press/v97/lu19a.html
PDF: http://proceedings.mlr.press/v97/lu19a/lu19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lu19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Songtao
family: Lu
- given: Mingyi
family: Hong
- given: Zhengdao
family: Wang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4134-4143
id: lu19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4134
lastpage: 4143
published: 2019-05-24 00:00:00 +0000
- title: 'Neurally-Guided Structure Inference'
abstract: 'Most structure inference methods either rely on exhaustive search or are purely data-driven. Exhaustive search robustly infers the structure of arbitrarily complex data, but it is slow. Data-driven methods allow efficient inference, but do not generalize when test data have more complex structures than training data. In this paper, we propose a hybrid inference algorithm, the Neurally-Guided Structure Inference (NG-SI), keeping the advantages of both search-based and data-driven methods. The key idea of NG-SI is to use a neural network to guide the hierarchical, layer-wise search over the compositional space of structures. We evaluate our algorithm on two representative structure inference tasks: probabilistic matrix decomposition and symbolic program parsing. It outperforms data-driven and search-based alternatives on both tasks.'
volume: 97
URL: https://proceedings.mlr.press/v97/lu19b.html
PDF: http://proceedings.mlr.press/v97/lu19b/lu19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lu19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sidi
family: Lu
- given: Jiayuan
family: Mao
- given: Joshua
family: Tenenbaum
- given: Jiajun
family: Wu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4144-4153
id: lu19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4144
lastpage: 4153
published: 2019-05-24 00:00:00 +0000
- title: 'Optimal Algorithms for Lipschitz Bandits with Heavy-tailed Rewards'
abstract: 'We study Lipschitz bandits, where a learner repeatedly plays one arm from an infinite arm set and then receives a stochastic reward whose expectation is a Lipschitz function of the chosen arm. Most of existing work assume the reward distributions are bounded or at least sub-Gaussian, and thus do not apply to heavy-tailed rewards arising in many real-world scenarios such as web advertising and financial markets. To address this limitation, in this paper we relax the assumption on rewards to allow arbitrary distributions that have finite $(1+\epsilon)$-th moments for some $\epsilon \in (0, 1]$, and propose algorithms that enjoy a sublinear regret of $\widetilde{O}(T^{(d_z\epsilon + 1)/(d_z \epsilon + \epsilon + 1)})$ where $T$ is the time horizon and $d_z$ is the zooming dimension. The key idea is to exploit the Lipschitz property of the expected reward function by adaptively discretizing the arm set, and employ upper confidence bound policies with robust mean estimators designed for heavy-tailed distributions. Furthermore, we provide a lower bound for Lipschitz bandits with heavy-tailed rewards, and show that our algorithms are optimal in terms of $T$. Finally, we conduct numerical experiments to demonstrate the effectiveness of our algorithms.'
volume: 97
URL: https://proceedings.mlr.press/v97/lu19c.html
PDF: http://proceedings.mlr.press/v97/lu19c/lu19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lu19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Shiyin
family: Lu
- given: Guanghui
family: Wang
- given: Yao
family: Hu
- given: Lijun
family: Zhang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4154-4163
id: lu19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4154
lastpage: 4163
published: 2019-05-24 00:00:00 +0000
- title: 'CoT: Cooperative Training for Generative Modeling of Discrete Data'
abstract: 'In this paper, we study the generative models of sequential discrete data. To tackle the exposure bias problem inherent in maximum likelihood estimation (MLE), generative adversarial networks (GANs) are introduced to penalize the unrealistic generated samples. To exploit the supervision signal from the discriminator, most previous models leverage REINFORCE to address the non-differentiable problem of sequential discrete data. However, because of the unstable property of the training signal during the dynamic process of adversarial training, the effectiveness of REINFORCE, in this case, is hardly guaranteed. To deal with such a problem, we propose a novel approach called Cooperative Training (CoT) to improve the training of sequence generative models. CoT transforms the min-max game of GANs into a joint maximization framework and manages to explicitly estimate and optimize Jensen-Shannon divergence. Moreover, CoT works without the necessity of pre-training via MLE, which is crucial to the success of previous methods. In the experiments, compared to existing state-of-the-art methods, CoT shows superior or at least competitive performance on sample quality, diversity, as well as training stability.'
volume: 97
URL: https://proceedings.mlr.press/v97/lu19d.html
PDF: http://proceedings.mlr.press/v97/lu19d/lu19d.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lu19d.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sidi
family: Lu
- given: Lantao
family: Yu
- given: Siyuan
family: Feng
- given: Yaoming
family: Zhu
- given: Weinan
family: Zhang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4164-4172
id: lu19d
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4164
lastpage: 4172
published: 2019-05-24 00:00:00 +0000
- title: 'Generalized Approximate Survey Propagation for High-Dimensional Estimation'
abstract: 'In Generalized Linear Estimation (GLE) problems, we seek to estimate a signal that is observed through a linear transform followed by a component-wise, possibly nonlinear and noisy, channel. In the Bayesian optimal setting, Generalized Approximate Message Passing (GAMP) is known to achieve optimal performance for GLE. However, its performance can significantly deteriorate whenever there is a mismatch between the assumed and the true generative model, a situation frequently encountered in practice. In this paper, we propose a new algorithm, named Generalized Approximate Survey Propagation (GASP), for solving GLE in the presence of prior or model misspecifications. As a prototypical example, we consider the phase retrieval problem, where we show that GASP outperforms the corresponding GAMP, reducing the reconstruction threshold and, for certain choices of its parameters, approaching Bayesian optimal performance. Furthermore, we present a set of state evolution equations that can precisely characterize the performance of GASP in the high-dimensional limit.'
volume: 97
URL: https://proceedings.mlr.press/v97/lucibello19a.html
PDF: http://proceedings.mlr.press/v97/lucibello19a/lucibello19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lucibello19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Carlo
family: Lucibello
- given: Luca
family: Saglietti
- given: Yue
family: Lu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4173-4182
id: lucibello19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4173
lastpage: 4182
published: 2019-05-24 00:00:00 +0000
- title: 'High-Fidelity Image Generation With Fewer Labels'
abstract: 'Deep generative models are becoming a cornerstone of modern machine learning. Recent work on conditional generative adversarial networks has shown that learning complex, high-dimensional distributions over natural images is within reach. While the latest models are able to generate high-fidelity, diverse natural images at high resolution, they rely on a vast quantity of labeled data. In this work we demonstrate how one can benefit from recent work on self- and semi-supervised learning to outperform the state of the art on both unsupervised ImageNet synthesis, as well as in the conditional setting. In particular, the proposed approach is able to match the sample quality (as measured by FID) of the current state-of-the-art conditional model BigGAN on ImageNet using only 10% of the labels and outperform it using 20% of the labels.'
volume: 97
URL: https://proceedings.mlr.press/v97/lucic19a.html
PDF: http://proceedings.mlr.press/v97/lucic19a/lucic19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-lucic19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Mario
family: Lučić
- given: Michael
family: Tschannen
- given: Marvin
family: Ritter
- given: Xiaohua
family: Zhai
- given: Olivier
family: Bachem
- given: Sylvain
family: Gelly
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4183-4192
id: lucic19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4183
lastpage: 4192
published: 2019-05-24 00:00:00 +0000
- title: 'Leveraging Low-Rank Relations Between Surrogate Tasks in Structured Prediction'
abstract: 'We study the interplay between surrogate methods for structured prediction and techniques from multitask learning designed to leverage relationships between surrogate outputs. We propose an efficient algorithm based on trace norm regularization which, differently from previous methods, does not require explicit knowledge of the coding/decoding functions of the surrogate framework. As a result, our algorithm can be applied to the broad class of problems in which the surrogate space is large or even infinite dimensional. We study excess risk bounds for trace norm regularized structured prediction proving the consistency and learning rates for our estimator. We also identify relevant regimes in which our approach can enjoy better generalization performance than previous methods. Numerical experiments on ranking problems indicate that enforcing low-rank relations among surrogate outputs may indeed provide a significant advantage in practice.'
volume: 97
URL: https://proceedings.mlr.press/v97/luise19a.html
PDF: http://proceedings.mlr.press/v97/luise19a/luise19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-luise19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Giulia
family: Luise
- given: Dimitrios
family: Stamos
- given: Massimiliano
family: Pontil
- given: Carlo
family: Ciliberto
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4193-4202
id: luise19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4193
lastpage: 4202
published: 2019-05-24 00:00:00 +0000
- title: 'Differentiable Dynamic Normalization for Learning Deep Representation'
abstract: 'This work presents Dynamic Normalization (DN), which is able to learn arbitrary normalization operations for different convolutional layers in a deep ConvNet. Unlike existing normalization approaches that predefined computations of the statistics (mean and variance), DN learns to estimate them. DN has several appealing benefits. First, it adapts to various networks, tasks, and batch sizes. Second, it can be easily implemented and trained in a differentiable end-to-end manner with merely small number of parameters. Third, its matrix formulation represents a wide range of normalization methods, shedding light on analyzing them theoretically. Extensive studies show that DN outperforms its counterparts in CIFAR10 and ImageNet.'
volume: 97
URL: https://proceedings.mlr.press/v97/luo19a.html
PDF: http://proceedings.mlr.press/v97/luo19a/luo19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-luo19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ping
family: Luo
- given: Peng
family: Zhanglin
- given: Shao
family: Wenqi
- given: Zhang
family: Ruimao
- given: Ren
family: Jiamin
- given: Wu
family: Lingyun
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4203-4211
id: luo19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4203
lastpage: 4211
published: 2019-05-24 00:00:00 +0000
- title: 'Disentangled Graph Convolutional Networks'
abstract: 'The formation of a real-world graph typically arises from the highly complex interaction of many latent factors. The existing deep learning methods for graph-structured data neglect the entanglement of the latent factors, rendering the learned representations non-robust and hardly explainable. However, learning representations that disentangle the latent factors poses great challenges and remains largely unexplored in the literature of graph neural networks. In this paper, we introduce the disentangled graph convolutional network (DisenGCN) to learn disentangled node representations. In particular, we propose a novel neighborhood routing mechanism, which is capable of dynamically identifying the latent factor that may have caused the edge between a node and one of its neighbors, and accordingly assigning the neighbor to a channel that extracts and convolutes features specific to that factor. We theoretically prove the convergence properties of the routing mechanism. Empirical results show that our proposed model can achieve significant performance gains, especially when the data demonstrate the existence of many entangled factors.'
volume: 97
URL: https://proceedings.mlr.press/v97/ma19a.html
PDF: http://proceedings.mlr.press/v97/ma19a/ma19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ma19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jianxin
family: Ma
- given: Peng
family: Cui
- given: Kun
family: Kuang
- given: Xin
family: Wang
- given: Wenwu
family: Zhu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4212-4221
id: ma19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4212
lastpage: 4221
published: 2019-05-24 00:00:00 +0000
- title: 'Variational Implicit Processes'
abstract: 'We introduce the implicit processes (IPs), a stochastic process that places implicitly defined multivariate distributions over any finite collections of random variables. IPs are therefore highly flexible implicit priors over *functions*, with examples including data simulators, Bayesian neural networks and non-linear transformations of stochastic processes. A novel and efficient approximate inference algorithm for IPs, namely the variational implicit processes (VIPs), is derived using generalised wake-sleep updates. This method returns simple update equations and allows scalable hyper-parameter learning with stochastic optimization. Experiments show that VIPs return better uncertainty estimates and lower errors over existing inference methods for challenging models such as Bayesian neural networks, and Gaussian processes.'
volume: 97
URL: https://proceedings.mlr.press/v97/ma19b.html
PDF: http://proceedings.mlr.press/v97/ma19b/ma19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ma19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Chao
family: Ma
- given: Yingzhen
family: Li
- given: Jose Miguel
family: Hernandez-Lobato
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4222-4233
id: ma19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4222
lastpage: 4233
published: 2019-05-24 00:00:00 +0000
- title: 'EDDI: Efficient Dynamic Discovery of High-Value Information with Partial VAE'
abstract: 'Many real-life decision making situations allow further relevant information to be acquired at a specific cost, for example, in assessing the health status of a patient we may decide to take additional measurements such as diagnostic tests or imaging scans before making a final assessment. Acquiring more relevant information enables better decision making, but may be costly. How can we trade off the desire to make good decisions by acquiring further information with the cost of performing that acquisition? To this end, we propose a principled framework, named *EDDI* (Efficient Dynamic Discovery of high-value Information), based on the theory of Bayesian experimental design. In EDDI, we propose a novel *partial variational autoencoder* (Partial VAE) to predict missing data entries problematically given any subset of the observed ones, and combine it with an acquisition function that maximizes expected information gain on a set of target variables. We show cost reduction at the same decision quality and improved decision quality at the same cost in multiple machine learning benchmarks and two real-world health-care applications.'
volume: 97
URL: https://proceedings.mlr.press/v97/ma19c.html
PDF: http://proceedings.mlr.press/v97/ma19c/ma19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ma19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Chao
family: Ma
- given: Sebastian
family: Tschiatschek
- given: Konstantina
family: Palla
- given: Jose Miguel
family: Hernandez-Lobato
- given: Sebastian
family: Nowozin
- given: Cheng
family: Zhang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4234-4243
id: ma19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4234
lastpage: 4243
published: 2019-05-24 00:00:00 +0000
- title: 'Bayesian leave-one-out cross-validation for large data'
abstract: 'Model inference, such as model comparison, model checking, and model selection, is an important part of model development. Leave-one-out cross-validation (LOO) is a general approach for assessing the generalizability of a model, but unfortunately, LOO does not scale well to large datasets. We propose a combination of using approximate inference techniques and probability-proportional-to-size-sampling (PPS) for fast LOO model evaluation for large datasets. We provide both theoretical and empirical results showing good properties for large data.'
volume: 97
URL: https://proceedings.mlr.press/v97/magnusson19a.html
PDF: http://proceedings.mlr.press/v97/magnusson19a/magnusson19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-magnusson19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Måns
family: Magnusson
- given: Michael
family: Andersen
- given: Johan
family: Jonasson
- given: Aki
family: Vehtari
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4244-4253
id: magnusson19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4244
lastpage: 4253
published: 2019-05-24 00:00:00 +0000
- title: 'Composable Core-sets for Determinant Maximization: A Simple Near-Optimal Algorithm'
abstract: '“Composable core-sets” are an efficient framework for solving optimization problems in massive data models. In this work, we consider efficient construction of composable core-sets for the determinant maximization problem. This can also be cast as the MAP inference task for “determinantal point processes", that have recently gained a lot of interest for modeling diversity and fairness. The problem was recently studied in \cite{indyk2018composable}, where they designed composable core-sets with the optimal approximation bound of $O(k)^k$. On the other hand, the more practical “Greedy" algorithm has been previously used in similar contexts. In this work, first we provide a theoretical approximation guarantee of $C^{k^2}$ for the Greedy algorithm in the context of composable core-sets; Further, we propose to use a “Local Search" based algorithm that while being still practical, achieves a nearly optimal approximation bound of $O(k)^{2k}$; Finally, we implement all three algorithms and show the effectiveness of our proposed algorithm on standard data sets.'
volume: 97
URL: https://proceedings.mlr.press/v97/mahabadi19a.html
PDF: http://proceedings.mlr.press/v97/mahabadi19a/mahabadi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mahabadi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sepideh
family: Mahabadi
- given: Piotr
family: Indyk
- given: Shayan Oveis
family: Gharan
- given: Alireza
family: Rezaei
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4254-4263
id: mahabadi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4254
lastpage: 4263
published: 2019-05-24 00:00:00 +0000
- title: 'Guided evolutionary strategies: augmenting random search with surrogate gradients'
abstract: 'Many applications in machine learning require optimizing a function whose true gradient is unknown or computationally expensive, but where surrogate gradient information, directions that may be correlated with the true gradient, is cheaply available. For example, this occurs when an approximate gradient is easier to compute than the full gradient (e.g. in meta-learning or unrolled optimization), or when a true gradient is intractable and is replaced with a surrogate (e.g. in reinforcement learning or training networks with discrete variables). We propose Guided Evolutionary Strategies (GES), a method for optimally using surrogate gradient directions to accelerate random search. GES defines a search distribution for evolutionary strategies that is elongated along a subspace spanned by the surrogate gradients and estimates a descent direction which can then be passed to a first-order optimizer. We analytically and numerically characterize the tradeoffs that result from tuning how strongly the search distribution is stretched along the guiding subspace and use this to derive a setting of the hyperparameters that works well across problems. We evaluate GES on several example problems, demonstrating an improvement over both standard evolutionary strategies and first-order methods that directly follow the surrogate gradient.'
volume: 97
URL: https://proceedings.mlr.press/v97/maheswaranathan19a.html
PDF: http://proceedings.mlr.press/v97/maheswaranathan19a/maheswaranathan19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-maheswaranathan19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Niru
family: Maheswaranathan
- given: Luke
family: Metz
- given: George
family: Tucker
- given: Dami
family: Choi
- given: Jascha
family: Sohl-Dickstein
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4264-4273
id: maheswaranathan19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4264
lastpage: 4273
published: 2019-05-24 00:00:00 +0000
- title: 'Data Poisoning Attacks in Multi-Party Learning'
abstract: 'In this work, we demonstrate universal multi-party poisoning attacks that adapt and apply to any multi-party learning process with arbitrary interaction pattern between the parties. More generally, we introduce and study $(k,p)$-poisoning attacks in which an adversary controls $k\in[m]$ of the parties, and for each corrupted party $P_i$, the adversary submits some poisoned data $T’_i$ on behalf of $P_i$ that is still "$(1-p)$-close" to the correct data $T_i$ (e.g., $1-p$ fraction of $T’_i$ is still honestly generated).We prove that for any "bad" property $B$ of the final trained hypothesis $h$ (e.g., $h$ failing on a particular test example or having "large" risk) that has an arbitrarily small constant probability of happening without the attack, there always is a $(k,p)$-poisoning attack that increases the probability of $B$ from $\mu$ to by $\mu^{1-p \cdot k/m} = \mu + \Omega(p \cdot k/m)$. Our attack only uses clean labels, and it is online, as it only knows the the data shared so far.'
volume: 97
URL: https://proceedings.mlr.press/v97/mahloujifar19a.html
PDF: http://proceedings.mlr.press/v97/mahloujifar19a/mahloujifar19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mahloujifar19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Saeed
family: Mahloujifar
- given: Mohammad
family: Mahmoody
- given: Ameer
family: Mohammed
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4274-4283
id: mahloujifar19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4274
lastpage: 4283
published: 2019-05-24 00:00:00 +0000
- title: 'Traditional and Heavy Tailed Self Regularization in Neural Network Models'
abstract: 'Random Matrix Theory (RMT) is applied to analyze the weight matrices of Deep Neural Networks (DNNs), including both production quality, pre-trained models such as AlexNet and Inception, and smaller models trained from scratch, such as LeNet5 and a miniature-AlexNet. Empirical and theoretical results clearly indicate that the empirical spectral density (ESD) of DNN layer matrices displays signatures of traditionally-regularized statistical models, even in the absence of exogenously specifying traditional forms of regularization, such as Dropout or Weight Norm constraints. Building on recent results in RMT, most notably its extension to Universality classes of Heavy-Tailed matrices, we develop a theory to identify *5+1 Phases of Training*, corresponding to increasing amounts of *Implicit Self-Regularization*. For smaller and/or older DNNs, this Implicit Self-Regularization is like traditional Tikhonov regularization, in that there is a “size scale” separating signal from noise. For state-of-the-art DNNs, however, we identify a novel form of *Heavy-Tailed Self-Regularization*, similar to the self-organization seen in the statistical physics of disordered systems. This implicit Self-Regularization can depend strongly on the many knobs of the training process. By exploiting the generalization gap phenomena, we demonstrate that we can cause a small model to exhibit all 5+1 phases of training simply by changing the batch size.'
volume: 97
URL: https://proceedings.mlr.press/v97/mahoney19a.html
PDF: http://proceedings.mlr.press/v97/mahoney19a/mahoney19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mahoney19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Michael
family: Mahoney
- given: Charles
family: Martin
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4284-4293
id: mahoney19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4284
lastpage: 4293
published: 2019-05-24 00:00:00 +0000
- title: 'Curvature-Exploiting Acceleration of Elastic Net Computations'
abstract: 'This paper introduces an efficient second-order method for solving the elastic net problem. Its key innovation is a computationally efficient technique for injecting curvature information in the optimization process which admits a strong theoretical performance guarantee. In particular, we show improved run time over popular first-order methods and quantify the speed-up in terms of statistical measures of the data matrix. The improved time complexity is the result of an extensive exploitation of the problem structure and a careful combination of second-order information, variance reduction techniques, and momentum acceleration. Beside theoretical speed-up, experimental results demonstrate great practical performance benefits of curvature information, especially for ill-conditioned data sets.'
volume: 97
URL: https://proceedings.mlr.press/v97/mai19a.html
PDF: http://proceedings.mlr.press/v97/mai19a/mai19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mai19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Vien
family: Mai
- given: Mikael
family: Johansson
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4294-4303
id: mai19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4294
lastpage: 4303
published: 2019-05-24 00:00:00 +0000
- title: 'Breaking the gridlock in Mixture-of-Experts: Consistent and Efficient Algorithms'
abstract: 'Mixture-of-Experts (MoE) is a widely popular model for ensemble learning and is a basic building block of highly successful modern neural networks as well as a component in Gated Recurrent Units (GRU) and Attention networks. However, present algorithms for learning MoE, including the EM algorithm and gradient descent, are known to get stuck in local optima. From a theoretical viewpoint, finding an efficient and provably consistent algorithm to learn the parameters remains a long standing open problem for more than two decades. In this paper, we introduce the first algorithm that learns the true parameters of a MoE model for a wide class of non-linearities with global consistency guarantees. While existing algorithms jointly or iteratively estimate the expert parameters and the gating parameters in the MoE, we propose a novel algorithm that breaks the deadlock and can directly estimate the expert parameters by sensing its echo in a carefully designed cross-moment tensor between the inputs and the output. Once the experts are known, the recovery of gating parameters still requires an EM algorithm; however, we show that the EM algorithm for this simplified problem, unlike the joint EM algorithm, converges to the true parameters. We empirically validate our algorithm on both the synthetic and real data sets in a variety of settings, and show superior performance to standard baselines.'
volume: 97
URL: https://proceedings.mlr.press/v97/makkuva19a.html
PDF: http://proceedings.mlr.press/v97/makkuva19a/makkuva19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-makkuva19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ashok
family: Makkuva
- given: Pramod
family: Viswanath
- given: Sreeram
family: Kannan
- given: Sewoong
family: Oh
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4304-4313
id: makkuva19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4304
lastpage: 4313
published: 2019-05-24 00:00:00 +0000
- title: 'Calibrated Model-Based Deep Reinforcement Learning'
abstract: 'Estimates of predictive uncertainty are important for accurate model-based planning and reinforcement learning. However, predictive uncertainties — especially ones derived from modern deep learning systems — can be inaccurate and impose a bottleneck on performance. This paper explores which uncertainties are needed for model-based reinforcement learning and argues that ideal uncertainties should be calibrated, i.e. their probabilities should match empirical frequencies of predicted events. We describe a simple way to augment any model-based reinforcement learning agent with a calibrated model and show that doing so consistently improves planning, sample complexity, and exploration. On the \textsc{HalfCheetah} MuJoCo task, our system achieves state-of-the-art performance using 50% fewer samples than the current leading approach. Our findings suggest that calibration can improve the performance of model-based reinforcement learning with minimal computational and implementation overhead.'
volume: 97
URL: https://proceedings.mlr.press/v97/malik19a.html
PDF: http://proceedings.mlr.press/v97/malik19a/malik19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-malik19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ali
family: Malik
- given: Volodymyr
family: Kuleshov
- given: Jiaming
family: Song
- given: Danny
family: Nemer
- given: Harlan
family: Seymour
- given: Stefano
family: Ermon
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4314-4323
id: malik19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4314
lastpage: 4323
published: 2019-05-24 00:00:00 +0000
- title: 'Learning from Delayed Outcomes via Proxies with Applications to Recommender Systems'
abstract: 'Predicting delayed outcomes is an important problem in recommender systems (e.g., if customers will finish reading an ebook). We formalize the problem as an adversarial, delayed online learning problem and consider how a proxy for the delayed outcome (e.g., if customers read a third of the book in 24 hours) can help minimize regret, even though the proxy is not available when making a prediction. Motivated by our regret analysis, we propose two neural network architectures: Factored Forecaster (FF) which is ideal if the proxy is informative of the outcome in hindsight, and Residual Factored Forecaster (RFF) that is robust to a non-informative proxy. Experiments on two real-world datasets for predicting human behavior show that RFF outperforms both FF and a direct forecaster that does not make use of the proxy. Our results suggest that exploiting proxies by factorization is a promising way to mitigate the impact of long delays in human-behavior prediction tasks.'
volume: 97
URL: https://proceedings.mlr.press/v97/mann19a.html
PDF: http://proceedings.mlr.press/v97/mann19a/mann19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mann19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Timothy Arthur
family: Mann
- given: Sven
family: Gowal
- given: Andras
family: Gyorgy
- given: Huiyi
family: Hu
- given: Ray
family: Jiang
- given: Balaji
family: Lakshminarayanan
- given: Prav
family: Srinivasan
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4324-4332
id: mann19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4324
lastpage: 4332
published: 2019-05-24 00:00:00 +0000
- title: 'Passed & Spurious: Descent Algorithms and Local Minima in Spiked Matrix-Tensor Models'
abstract: 'In this work we analyse quantitatively the interplay between the loss landscape and performance of descent algorithms in a prototypical inference problem, the spiked matrix-tensor model. We study a loss function that is the negative log-likelihood of the model. We analyse the number of local minima at a fixed distance from the signal/spike with the Kac-Rice formula, and locate trivialization of the landscape at large signal-to-noise ratios. We evaluate analytically the performance of a gradient flow algorithm using integro-differential PDEs as developed in physics of disordered systems for the Langevin dynamics. We analyze the performance of an approximate message passing algorithm estimating the maximum likelihood configuration via its state evolution. We conclude by comparing the above results: while we observe a drastic slow down of the gradient flow dynamics even in the region where the landscape is trivial, both the analyzed algorithms are shown to perform well even in the part of the region of parameters where spurious local minima are present.'
volume: 97
URL: https://proceedings.mlr.press/v97/mannelli19a.html
PDF: http://proceedings.mlr.press/v97/mannelli19a/mannelli19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mannelli19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Stefano Sarao
family: Mannelli
- given: Florent
family: Krzakala
- given: Pierfrancesco
family: Urbani
- given: Lenka
family: Zdeborova
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4333-4342
id: mannelli19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4333
lastpage: 4342
published: 2019-05-24 00:00:00 +0000
- title: 'A Baseline for Any Order Gradient Estimation in Stochastic Computation Graphs'
abstract: 'By enabling correct differentiation in Stochastic Computation Graphs (SCGs), the infinitely differentiable Monte-Carlo estimator (DiCE) can generate correct estimates for the higher order gradients that arise in, e.g., multi-agent reinforcement learning and meta-learning. However, the baseline term in DiCE that serves as a control variate for reducing variance applies only to first order gradient estimation, limiting the utility of higher-order gradient estimates. To improve the sample efficiency of DiCE, we propose a new baseline term for higher order gradient estimation. This term may be easily included in the objective, and produces unbiased variance-reduced estimators under (automatic) differentiation, without affecting the estimate of the objective itself or of the first order gradient estimate. It reuses the same baseline function (e.g., the state-value function in reinforcement learning) already used for the first order baseline. We provide theoretical analysis and numerical evaluations of this new baseline, which demonstrate that it can dramatically reduce the variance of DiCE’s second order gradient estimators and also show empirically that it reduces the variance of third and fourth order gradients. This computational tool can be easily used to estimate higher order gradients with unprecedented efficiency and simplicity wherever automatic differentiation is utilised, and it has the potential to unlock applications of higher order gradients in reinforcement learning and meta-learning.'
volume: 97
URL: https://proceedings.mlr.press/v97/mao19a.html
PDF: http://proceedings.mlr.press/v97/mao19a/mao19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mao19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jingkai
family: Mao
- given: Jakob
family: Foerster
- given: Tim
family: Rocktäschel
- given: Maruan
family: Al-Shedivat
- given: Gregory
family: Farquhar
- given: Shimon
family: Whiteson
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4343-4351
id: mao19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4343
lastpage: 4351
published: 2019-05-24 00:00:00 +0000
- title: 'Adversarial Generation of Time-Frequency Features with application in audio synthesis'
abstract: 'Time-frequency (TF) representations provide powerful and intuitive features for the analysis of time series such as audio. But still, generative modeling of audio in the TF domain is a subtle matter. Consequently, neural audio synthesis widely relies on directly modeling the waveform and previous attempts at unconditionally synthesizing audio from neurally generated invertible TF features still struggle to produce audio at satisfying quality. In this article, focusing on the short-time Fourier transform, we discuss the challenges that arise in audio synthesis based on generated invertible TF features and how to overcome them. We demonstrate the potential of deliberate generative TF modeling by training a generative adversarial network (GAN) on short-time Fourier features. We show that by applying our guidelines, our TF-based network was able to outperform a state-of-the-art GAN generating waveforms directly, despite the similar architecture in the two networks.'
volume: 97
URL: https://proceedings.mlr.press/v97/marafioti19a.html
PDF: http://proceedings.mlr.press/v97/marafioti19a/marafioti19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-marafioti19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Andrés
family: Marafioti
- given: Nathanaël
family: Perraudin
- given: Nicki
family: Holighaus
- given: Piotr
family: Majdak
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4352-4362
id: marafioti19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4352
lastpage: 4362
published: 2019-05-24 00:00:00 +0000
- title: 'On the Universality of Invariant Networks'
abstract: 'Constraining linear layers in neural networks to respect symmetry transformations from a group $G$ is a common design principle for invariant networks that has found many applications in machine learning. In this paper, we consider a fundamental question that has received very little attention to date: Can these networks approximate any (continuous) invariant function? We tackle the rather general case where $G\leq S_n$ (an arbitrary subgroup of the symmetric group) that acts on $\R^n$ by permuting coordinates. This setting includes several recent popular invariant networks. We present two main results: First, $G$-invariant networks are universal if high-order tensors are allowed. Second, there are groups $G$ for which higher-order tensors are unavoidable for obtaining universality. $G$-invariant networks consisting of only first-order tensors are of special interest due to their practical value. We conclude the paper by proving a necessary condition for the universality of $G$-invariant networks that incorporate only first-order tensors. Lastly, we propose a conjecture stating that this condition is also sufficient.'
volume: 97
URL: https://proceedings.mlr.press/v97/maron19a.html
PDF: http://proceedings.mlr.press/v97/maron19a/maron19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-maron19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Haggai
family: Maron
- given: Ethan
family: Fetaya
- given: Nimrod
family: Segol
- given: Yaron
family: Lipman
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4363-4371
id: maron19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4363
lastpage: 4371
published: 2019-05-24 00:00:00 +0000
- title: 'Decomposing feature-level variation with Covariate Gaussian Process Latent Variable Models'
abstract: 'The interpretation of complex high-dimensional data typically requires the use of dimensionality reduction techniques to extract explanatory low-dimensional representations. However, in many real-world problems these representations may not be sufficient to aid interpretation on their own, and it would be desirable to interpret the model in terms of the original features themselves. Our goal is to characterise how feature-level variation depends on latent low-dimensional representations, external covariates, and non-linear interactions between the two. In this paper, we propose to achieve this through a structured kernel decomposition in a hybrid Gaussian Process model which we call the Covariate Gaussian Process Latent Variable Model (c-GPLVM). We demonstrate the utility of our model on simulated examples and applications in disease progression modelling from high-dimensional gene expression data in the presence of additional phenotypes. In each setting we show how the c-GPLVM can extract low-dimensional structures from high-dimensional data sets whilst allowing a breakdown of feature-level variability that is not present in other commonly used dimensionality reduction approaches.'
volume: 97
URL: https://proceedings.mlr.press/v97/martens19a.html
PDF: http://proceedings.mlr.press/v97/martens19a/martens19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-martens19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Kaspar
family: Märtens
- given: Kieran
family: Campbell
- given: Christopher
family: Yau
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4372-4381
id: martens19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4372
lastpage: 4381
published: 2019-05-24 00:00:00 +0000
- title: 'Fairness-Aware Learning for Continuous Attributes and Treatments'
abstract: 'We address the problem of algorithmic fairness: ensuring that the outcome of a classifier is not biased towards certain values of sensitive variables such as age, race or gender. As common fairness metrics can be expressed as measures of (conditional) independence between variables, we propose to use the Rényi maximum correlation coefficient to generalize fairness measurement to continuous variables. We exploit Witsenhausen’s characterization of the Rényi correlation coefficient to propose a differentiable implementation linked to $f$-divergences. This allows us to generalize fairness-aware learning to continuous variables by using a penalty that upper bounds this coefficient. Theses allows fairness to be extented to variables such as mixed ethnic groups or financial status without thresholds effects. This penalty can be estimated on mini-batches allowing to use deep nets. Experiments show favorable comparisons to state of the art on binary variables and prove the ability to protect continuous ones'
volume: 97
URL: https://proceedings.mlr.press/v97/mary19a.html
PDF: http://proceedings.mlr.press/v97/mary19a/mary19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mary19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jeremie
family: Mary
- given: Clément
family: Calauzènes
- given: Noureddine El
family: Karoui
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4382-4391
id: mary19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4382
lastpage: 4391
published: 2019-05-24 00:00:00 +0000
- title: 'Optimal Minimal Margin Maximization with Boosting'
abstract: 'Boosting algorithms iteratively produce linear combinations of more and more base hypotheses and it has been observed experimentally that the generalization error keeps improving even after achieving zero training error. One popular explanation attributes this to improvements in margins. A common goal in a long line of research, is to obtain large margins using as few base hypotheses as possible, culminating with the AdaBoostV algorithm by R{ä}tsch and Warmuth [JMLR’05]. The AdaBoostV algorithm was later conjectured to yield an optimal trade-off between number of hypotheses trained and the minimal margin over all training points (Nie, Warmuth, Vishwanathan and Zhang [JMLR’13]). Our main contribution is a new algorithm refuting this conjecture. Furthermore, we prove a lower bound which implies that our new algorithm is optimal.'
volume: 97
URL: https://proceedings.mlr.press/v97/mathiasen19a.html
PDF: http://proceedings.mlr.press/v97/mathiasen19a/mathiasen19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mathiasen19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alexander
family: Mathiasen
- given: Kasper Green
family: Larsen
- given: Allan
family: Grønlund
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4392-4401
id: mathiasen19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4392
lastpage: 4401
published: 2019-05-24 00:00:00 +0000
- title: 'Disentangling Disentanglement in Variational Autoencoders'
abstract: 'We develop a generalisation of disentanglement in variational autoencoders (VAEs)—decomposition of the latent representation—characterising it as the fulfilment of two factors: a) the latent encodings of the data having an appropriate level of overlap, and b) the aggregate encoding of the data conforming to a desired structure, represented through the prior. Decomposition permits disentanglement, i.e. explicit independence between latents, as a special case, but also allows for a much richer class of properties to be imposed on the learnt representation, such as sparsity, clustering, independent subspaces, or even intricate hierarchical dependency relationships. We show that the $\beta$-VAE varies from the standard VAE predominantly in its control of latent overlap and that for the standard choice of an isotropic Gaussian prior, its objective is invariant to rotations of the latent representation. Viewed from the decomposition perspective, breaking this invariance with simple manipulations of the prior can yield better disentanglement with little or no detriment to reconstructions. We further demonstrate how other choices of prior can assist in producing different decompositions and introduce an alternative training objective that allows the control of both decomposition factors in a principled manner.'
volume: 97
URL: https://proceedings.mlr.press/v97/mathieu19a.html
PDF: http://proceedings.mlr.press/v97/mathieu19a/mathieu19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mathieu19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Emile
family: Mathieu
- given: Tom
family: Rainforth
- given: N
family: Siddharth
- given: Yee Whye
family: Teh
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4402-4412
id: mathieu19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4402
lastpage: 4412
published: 2019-05-24 00:00:00 +0000
- title: 'MIWAE: Deep Generative Modelling and Imputation of Incomplete Data Sets'
abstract: 'We consider the problem of handling missing data with deep latent variable models (DLVMs). First, we present a simple technique to train DLVMs when the training set contains missing-at-random data. Our approach, called MIWAE, is based on the importance-weighted autoencoder (IWAE), and maximises a potentially tight lower bound of the log-likelihood of the observed data. Compared to the original IWAE, our algorithm does not induce any additional computational overhead due to the missing data. We also develop Monte Carlo techniques for single and multiple imputation using a DLVM trained on an incomplete data set. We illustrate our approach by training a convolutional DLVM on incomplete static binarisations of MNIST. Moreover, on various continuous data sets, we show that MIWAE provides extremely accurate single imputations, and is highly competitive with state-of-the-art methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/mattei19a.html
PDF: http://proceedings.mlr.press/v97/mattei19a/mattei19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mattei19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Pierre-Alexandre
family: Mattei
- given: Jes
family: Frellsen
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4413-4423
id: mattei19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4413
lastpage: 4423
published: 2019-05-24 00:00:00 +0000
- title: 'Distributional Reinforcement Learning for Efficient Exploration'
abstract: 'In distributional reinforcement learning (RL), the estimated distribution of value functions model both the parametric and intrinsic uncertainties. We propose a novel and efficient exploration method for deep RL that has two components. The first is a decaying schedule to suppress the intrinsic uncertainty. The second is an exploration bonus calculated from the upper quantiles of the learned distribution. In Atari 2600 games, our method achieves 483 % average gain across 49 games in cumulative rewards over QR-DQN. We also compared our algorithm with QR-DQN in a challenging 3D driving simulator (CARLA). Results show that our algorithm achieves nearoptimal safety rewards twice faster than QRDQN.'
volume: 97
URL: https://proceedings.mlr.press/v97/mavrin19a.html
PDF: http://proceedings.mlr.press/v97/mavrin19a/mavrin19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mavrin19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Borislav
family: Mavrin
- given: Hengshuai
family: Yao
- given: Linglong
family: Kong
- given: Kaiwen
family: Wu
- given: Yaoliang
family: Yu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4424-4434
id: mavrin19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4424
lastpage: 4434
published: 2019-05-24 00:00:00 +0000
- title: 'Graphical-model based estimation and inference for differential privacy'
abstract: 'Many privacy mechanisms reveal high-level information about a data distribution through noisy measurements. It is common to use this information to estimate the answers to new queries. In this work, we provide an approach to solve this estimation problem efficiently using graphical models, which is particularly effective when the distribution is high-dimensional but the measurements are over low-dimensional marginals. We show that our approach is far more efficient than existing estimation techniques from the privacy literature and that it can improve the accuracy and scalability of many state-of-the-art mechanisms.'
volume: 97
URL: https://proceedings.mlr.press/v97/mckenna19a.html
PDF: http://proceedings.mlr.press/v97/mckenna19a/mckenna19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mckenna19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ryan
family: Mckenna
- given: Daniel
family: Sheldon
- given: Gerome
family: Miklau
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4435-4444
id: mckenna19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4435
lastpage: 4444
published: 2019-05-24 00:00:00 +0000
- title: 'Efficient Amortised Bayesian Inference for Hierarchical and Nonlinear Dynamical Systems'
abstract: 'We introduce a flexible, scalable Bayesian inference framework for nonlinear dynamical systems characterised by distinct and hierarchical variability at the individual, group, and population levels. Our model class is a generalisation of nonlinear mixed-effects (NLME) dynamical systems, the statistical workhorse for many experimental sciences. We cast parameter inference as stochastic optimisation of an end-to-end differentiable, block-conditional variational autoencoder. We specify the dynamics of the data-generating process as an ordinary differential equation (ODE) such that both the ODE and its solver are fully differentiable. This model class is highly flexible: the ODE right-hand sides can be a mixture of user-prescribed or "white-box" sub-components and neural network or "black-box" sub-components. Using stochastic optimisation, our amortised inference algorithm could seamlessly scale up to massive data collection pipelines (common in labs with robotic automation). Finally, our framework supports interpretability with respect to the underlying dynamics, as well as predictive generalization to unseen combinations of group components (also called “zero-shot" learning). We empirically validate our method by predicting the dynamic behaviour of bacteria that were genetically engineered to function as biosensors.'
volume: 97
URL: https://proceedings.mlr.press/v97/meeds19a.html
PDF: http://proceedings.mlr.press/v97/meeds19a/meeds19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-meeds19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Geoffrey
family: Roeder
- given: Paul
family: Grant
- given: Andrew
family: Phillips
- given: Neil
family: Dalchau
- given: Edward
family: Meeds
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4445-4455
id: meeds19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4445
lastpage: 4455
published: 2019-05-24 00:00:00 +0000
- title: 'Toward Controlling Discrimination in Online Ad Auctions'
abstract: 'Online advertising platforms are thriving due to the customizable audiences they offer advertisers. However, recent studies show that advertisements can be discriminatory with respect to the gender or race of the audience that sees the ad, and may inadvertently cross ethical and/or legal boundaries. To prevent this, we propose a constrained ad auction framework that maximizes the platform’s revenue conditioned on ensuring that the audience seeing an advertiser’s ad is distributed appropriately across sensitive types such as gender or race. Building upon Myerson’s classic work, we first present an optimal auction mechanism for a large class of fairness constraints. Finding the parameters of this optimal auction, however, turns out to be a non-convex problem. We show that this non-convex problem can be reformulated as a more structured non-convex problem with no saddle points or local-maxima; this allows us to develop a gradient-descent-based algorithm to solve it. Our empirical results on the A1 Yahoo! dataset demonstrate that our algorithm can obtain uniform coverage across different user types for each advertiser at a minor loss to the revenue of the platform, and a small change to the size of the audience each advertiser reaches.'
volume: 97
URL: https://proceedings.mlr.press/v97/mehrotra19a.html
PDF: http://proceedings.mlr.press/v97/mehrotra19a/mehrotra19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mehrotra19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Elisa
family: Celis
- given: Anay
family: Mehrotra
- given: Nisheeth
family: Vishnoi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4456-4465
id: mehrotra19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4456
lastpage: 4465
published: 2019-05-24 00:00:00 +0000
- title: 'Stochastic Blockmodels meet Graph Neural Networks'
abstract: 'Stochastic blockmodels (SBM) and their variants, $e.g.$, mixed-membership and overlapping stochastic blockmodels, are latent variable based generative models for graphs. They have proven to be successful for various tasks, such as discovering the community structure and link prediction on graph-structured data. Recently, graph neural networks, $e.g.$, graph convolutional networks, have also emerged as a promising approach to learn powerful representations (embeddings) for the nodes in the graph, by exploiting graph properties such as locality and invariance. In this work, we unify these two directions by developing a *sparse* variational autoencoder for graphs, that retains the interpretability of SBMs, while also enjoying the excellent predictive performance of graph neural nets. Moreover, our framework is accompanied by a fast recognition model that enables fast inference of the node embeddings (which are of independent interest for inference in SBM and its variants). Although we develop this framework for a particular type of SBM, namely the *overlapping* stochastic blockmodel, the proposed framework can be adapted readily for other types of SBMs. Experimental results on several benchmarks demonstrate encouraging results on link prediction while learning an interpretable latent structure that can be used for community discovery.'
volume: 97
URL: https://proceedings.mlr.press/v97/mehta19a.html
PDF: http://proceedings.mlr.press/v97/mehta19a/mehta19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mehta19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Nikhil
family: Mehta
- given: Lawrence Carin
family: Duke
- given: Piyush
family: Rai
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4466-4474
id: mehta19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4466
lastpage: 4474
published: 2019-05-24 00:00:00 +0000
- title: 'Imputing Missing Events in Continuous-Time Event Streams'
abstract: 'Events in the world may be caused by other, unobserved events. We consider sequences of events in continuous time. Given a probability model of complete sequences, we propose particle smoothing—a form of sequential importance sampling—to impute the missing events in an incomplete sequence. We develop a trainable family of proposal distributions based on a type of bidirectional continuous-time LSTM: Bidirectionality lets the proposals condition on future observations, not just on the past as in particle filtering. Our method can sample an ensemble of possible complete sequences (particles), from which we form a single consensus prediction that has low Bayes risk under our chosen loss metric. We experiment in multiple synthetic and real domains, using different missingness mechanisms, and modeling the complete sequences in each domain with a neural Hawkes process (Mei & Eisner 2017). On held-out incomplete sequences, our method is effective at inferring the ground-truth unobserved events, with particle smoothing consistently improving upon particle filtering.'
volume: 97
URL: https://proceedings.mlr.press/v97/mei19a.html
PDF: http://proceedings.mlr.press/v97/mei19a/mei19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mei19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Hongyuan
family: Mei
- given: Guanghui
family: Qin
- given: Jason
family: Eisner
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4475-4485
id: mei19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4475
lastpage: 4485
published: 2019-05-24 00:00:00 +0000
- title: 'Same, Same But Different: Recovering Neural Network Quantization Error Through Weight Factorization'
abstract: 'Quantization of neural networks has become common practice, driven by the need for efficient implementations of deep neural networks on embedded devices. In this paper, we exploit an oft-overlooked degree of freedom in most networks - for a given layer, individual output channels can be scaled by any factor provided that the corresponding weights of the next layer are inversely scaled. Therefore, a given network has many factorizations which change the weights of the network without changing its function. We present a conceptually simple and easy to implement method that uses this property and show that proper factorizations significantly decrease the degradation caused by quantization. We show improvement on a wide variety of networks and achieve state-of-the-art degradation results for MobileNets. While our focus is on quantization, this type of factorization is applicable to other domains such as network-pruning, neural nets regularization and network interpretability.'
volume: 97
URL: https://proceedings.mlr.press/v97/meller19a.html
PDF: http://proceedings.mlr.press/v97/meller19a/meller19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-meller19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Eldad
family: Meller
- given: Alexander
family: Finkelstein
- given: Uri
family: Almog
- given: Mark
family: Grobman
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4486-4495
id: meller19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4486
lastpage: 4495
published: 2019-05-24 00:00:00 +0000
- title: 'The Wasserstein Transform'
abstract: 'We introduce the Wasserstein transform, a method for enhancing and denoising datasets defined on general metric spaces. The construction draws inspiration from Optimal Transportation ideas. We establish the stability of our method under data perturbation and, when the dataset is assumed to be Euclidean, we also exhibit a precise connection between the Wasserstein transform and the mean shift family of algorithms. We then use this connection to prove that mean shift also inherits stability under perturbations. We study the performance of the Wasserstein transform method on different datasets as a preprocessing step prior to clustering and classification tasks.'
volume: 97
URL: https://proceedings.mlr.press/v97/memoli19a.html
PDF: http://proceedings.mlr.press/v97/memoli19a/memoli19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-memoli19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Facundo
family: Memoli
- given: Zane
family: Smith
- given: Zhengchao
family: Wan
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4496-4504
id: memoli19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4496
lastpage: 4504
published: 2019-05-24 00:00:00 +0000
- title: 'Ithemal: Accurate, Portable and Fast Basic Block Throughput Estimation using Deep Neural Networks'
abstract: 'Predicting the number of clock cycles a processor takes to execute a block of assembly instructions in steady state (the throughput) is important for both compiler designers and performance engineers. Building an analytical model to do so is especially complicated in modern x86-64 Complex Instruction Set Computer (CISC) machines with sophisticated processor microarchitectures in that it is tedious, error prone, and must be performed from scratch for each processor generation. In this paper we present Ithemal, the first tool which learns to predict the throughput of a set of instructions. Ithemal uses a hierarchical LSTM–based approach to predict throughput based on the opcodes and operands of instructions in a basic block. We show that Ithemal is more accurate than state-of-the-art hand-written tools currently used in compiler backends and static machine code analyzers. In particular, our model has less than half the error of state-of-the-art analytical models (LLVM’s llvm-mca and Intel’s IACA). Ithemal is also able to predict these throughput values just as fast as the aforementioned tools, and is easily ported across a variety of processor microarchitectures with minimal developer effort.'
volume: 97
URL: https://proceedings.mlr.press/v97/mendis19a.html
PDF: http://proceedings.mlr.press/v97/mendis19a/mendis19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mendis19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Charith
family: Mendis
- given: Alex
family: Renda
- given: Dr.Saman
family: Amarasinghe
- given: Michael
family: Carbin
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4505-4515
id: mendis19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4505
lastpage: 4515
published: 2019-05-24 00:00:00 +0000
- title: 'Geometric Losses for Distributional Learning'
abstract: 'Building upon recent advances in entropy-regularized optimal transport, and upon Fenchel duality between measures and continuous functions, we propose a generalization of the logistic loss that incorporates a metric or cost between classes. Unlike previous attempts to use optimal transport distances for learning, our loss results in unconstrained convex objective functions, supports infinite (or very large) class spaces, and naturally defines a geometric generalization of the softmax operator. The geometric properties of this loss make it suitable for predicting sparse and singular distributions, for instance supported on curves or hyper-surfaces. We study the theoretical properties of our loss and showcase its effectiveness on two applications: ordinal regression and drawing generation.'
volume: 97
URL: https://proceedings.mlr.press/v97/mensch19a.html
PDF: http://proceedings.mlr.press/v97/mensch19a/mensch19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mensch19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Arthur
family: Mensch
- given: Mathieu
family: Blondel
- given: Gabriel
family: Peyré
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4516-4525
id: mensch19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4516
lastpage: 4525
published: 2019-05-24 00:00:00 +0000
- title: 'Spectral Clustering of Signed Graphs via Matrix Power Means'
abstract: 'Signed graphs encode positive (attractive) and negative (repulsive) relations between nodes. We extend spectral clustering to signed graphs via the one-parameter family of Signed Power Mean Laplacians, defined as the matrix power mean of normalized standard and signless Laplacians of positive and negative edges. We provide a thorough analysis of the proposed approach in the setting of a general Stochastic Block Model that includes models such as the Labeled Stochastic Block Model and the Censored Block Model. We show that in expectation the signed power mean Laplacian captures the ground truth clusters under reasonable settings where state-of-the-art approaches fail. Moreover, we prove that the eigenvalues and eigenvector of the signed power mean Laplacian concentrate around their expectation under reasonable conditions in the general Stochastic Block Model. Extensive experiments on random graphs and real world datasets confirm the theoretically predicted behaviour of the signed power mean Laplacian and show that it compares favourably with state-of-the-art methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/mercado19a.html
PDF: http://proceedings.mlr.press/v97/mercado19a/mercado19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mercado19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Pedro
family: Mercado
- given: Francesco
family: Tudisco
- given: Matthias
family: Hein
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4526-4536
id: mercado19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4526
lastpage: 4536
published: 2019-05-24 00:00:00 +0000
- title: 'Simple Stochastic Gradient Methods for Non-Smooth Non-Convex Regularized Optimization'
abstract: 'Our work focuses on stochastic gradient methods for optimizing a smooth non-convex loss function with a non-smooth non-convex regularizer. Research on this class of problem is quite limited, and until recently no non-asymptotic convergence results have been reported. We present two simple stochastic gradient algorithms, for finite-sum and general stochastic optimization problems, which have superior convergence complexities compared to the current state-of-the-art. We also compare our algorithms’ performance in practice for empirical risk minimization.'
volume: 97
URL: https://proceedings.mlr.press/v97/metel19a.html
PDF: http://proceedings.mlr.press/v97/metel19a/metel19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-metel19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Michael
family: Metel
- given: Akiko
family: Takeda
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4537-4545
id: metel19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4537
lastpage: 4545
published: 2019-05-24 00:00:00 +0000
- title: 'Reinforcement Learning in Configurable Continuous Environments'
abstract: 'Configurable Markov Decision Processes (Conf-MDPs) have been recently introduced as an extension of the usual MDP model to account for the possibility of configuring the environment to improve the agent’s performance. Currently, there is still no suitable algorithm to solve the learning problem for real-world Conf-MDPs. In this paper, we fill this gap by proposing a trust-region method, Relative Entropy Model Policy Search (REMPS), able to learn both the policy and the MDP configuration in continuous domains without requiring the knowledge of the true model of the environment. After introducing our approach and providing a finite-sample analysis, we empirically evaluate REMPS on both benchmark and realistic environments by comparing our results with those of the gradient methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/metelli19a.html
PDF: http://proceedings.mlr.press/v97/metelli19a/metelli19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-metelli19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alberto Maria
family: Metelli
- given: Emanuele
family: Ghelfi
- given: Marcello
family: Restelli
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4546-4555
id: metelli19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4546
lastpage: 4555
published: 2019-05-24 00:00:00 +0000
- title: 'Understanding and correcting pathologies in the training of learned optimizers'
abstract: 'Deep learning has shown that learned functions can dramatically outperform hand-designed functions on perceptual tasks. Analogously, this suggests that learned optimizers may similarly outperform current hand-designed optimizers, especially for specific problems. However, learned optimizers are notoriously difficult to train and have yet to demonstrate wall-clock speedups over hand-designed optimizers, and thus are rarely used in practice. Typically, learned optimizers are trained by truncated backpropagation through an unrolled optimization process. The resulting gradients are either strongly biased (for short truncations) or have exploding norm (for long truncations). In this work we propose a training scheme which overcomes both of these difficulties, by dynamically weighting two unbiased gradient estimators for a variational loss on optimizer performance. This allows us to train neural networks to perform optimization of a specific task faster than tuned first-order methods. Moreover, by training the optimizer against validation loss (as opposed to training loss), we are able to learn optimizers that train networks to generalize better than first order methods. We demonstrate these results on problems where our learned optimizer trains convolutional networks faster in wall-clock time compared to tuned first-order methods and with an improvement in test loss.'
volume: 97
URL: https://proceedings.mlr.press/v97/metz19a.html
PDF: http://proceedings.mlr.press/v97/metz19a/metz19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-metz19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Luke
family: Metz
- given: Niru
family: Maheswaranathan
- given: Jeremy
family: Nixon
- given: Daniel
family: Freeman
- given: Jascha
family: Sohl-Dickstein
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4556-4565
id: metz19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4556
lastpage: 4565
published: 2019-05-24 00:00:00 +0000
- title: 'Optimality Implies Kernel Sum Classifiers are Statistically Efficient'
abstract: 'We propose a novel combination of optimization tools with learning theory bounds in order to analyze the sample complexity of optimal kernel sum classifiers. This contrasts the typical learning theoretic results which hold for all (potentially suboptimal) classifiers. Our work also justifies assumptions made in prior work on multiple kernel learning. As a byproduct of our analysis, we also provide a new form of Rademacher complexity for hypothesis classes containing only optimal classifiers.'
volume: 97
URL: https://proceedings.mlr.press/v97/meyer19a.html
PDF: http://proceedings.mlr.press/v97/meyer19a/meyer19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-meyer19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Raphael
family: Meyer
- given: Jean
family: Honorio
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4566-4574
id: meyer19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4566
lastpage: 4574
published: 2019-05-24 00:00:00 +0000
- title: 'On Dropout and Nuclear Norm Regularization'
abstract: 'We give a formal and complete characterization of the explicit regularizer induced by dropout in deep linear networks with squared loss. We show that (a) the explicit regularizer is composed of an $\ell_2$-path regularizer and other terms that are also re-scaling invariant, (b) the convex envelope of the induced regularizer is the squared nuclear norm of the network map, and (c) for a sufficiently large dropout rate, we characterize the global optima of the dropout objective. We validate our theoretical findings with empirical results.'
volume: 97
URL: https://proceedings.mlr.press/v97/mianjy19a.html
PDF: http://proceedings.mlr.press/v97/mianjy19a/mianjy19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mianjy19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Poorya
family: Mianjy
- given: Raman
family: Arora
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4575-4584
id: mianjy19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4575
lastpage: 4584
published: 2019-05-24 00:00:00 +0000
- title: 'Discriminative Regularization for Latent Variable Models with Applications to Electrocardiography'
abstract: 'Generative models often use latent variables to represent structured variation in high-dimensional data, such as images and medical waveforms. However, these latent variables may ignore subtle, yet meaningful features in the data. Some features may predict an outcome of interest (e.g. heart attack) but account for only a small fraction of variation in the data. We propose a generative model training objective that uses a black-box discriminative model as a regularizer to learn representations that preserve this predictive variation. With these discriminatively regularized latent variable models, we visualize and measure variation in the data that influence a black-box predictive model, enabling an expert to better understand each prediction. With this technique, we study models that use electrocardiograms to predict outcomes of clinical interest. We measure our approach on synthetic and real data with statistical summaries and an experiment carried out by a physician.'
volume: 97
URL: https://proceedings.mlr.press/v97/miller19a.html
PDF: http://proceedings.mlr.press/v97/miller19a/miller19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-miller19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Andrew
family: Miller
- given: Ziad
family: Obermeyer
- given: John
family: Cunningham
- given: Sendhil
family: Mullainathan
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4585-4594
id: miller19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4585
lastpage: 4594
published: 2019-05-24 00:00:00 +0000
- title: 'Formal Privacy for Functional Data with Gaussian Perturbations'
abstract: 'Motivated by the rapid rise in statistical tools in *Functional Data Analysis*, we consider the Gaussian mechanism for achieving differential privacy (DP) with parameter estimates taking values in a, potentially infinite-dimensional, separable Banach space. Using classic results from probability theory, we show how densities over function spaces can be utilized to achieve the desired DP bounds. This extends prior results of Hall et al (2013) to a much broader class of statistical estimates and summaries, including “path level" summaries, nonlinear functionals, and full function releases. By focusing on Banach spaces, we provide a deeper picture of the challenges for privacy with complex data, especially the role regularization plays in balancing utility and privacy. Using an application to penalized smoothing, we highlight this balance in the context of mean function estimation. Simulations and an application to {diffusion tensor imaging} are briefly presented, with extensive additions included in a supplement.'
volume: 97
URL: https://proceedings.mlr.press/v97/mirshani19a.html
PDF: http://proceedings.mlr.press/v97/mirshani19a/mirshani19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mirshani19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ardalan
family: Mirshani
- given: Matthew
family: Reimherr
- given: Aleksandra
family: Slavković
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4595-4604
id: mirshani19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4595
lastpage: 4604
published: 2019-05-24 00:00:00 +0000
- title: 'Co-manifold learning with missing data'
abstract: 'Representation learning is typically applied to only one mode of a data matrix, either its rows or columns. Yet in many applications, there is an underlying geometry to both the rows and the columns. We propose utilizing this coupled structure to perform co-manifold learning: uncovering the underlying geometry of both the rows and the columns of a given matrix, where we focus on a missing data setting. Our unsupervised approach consists of three components. We first solve a family of optimization problems to estimate a complete matrix at multiple scales of smoothness. We then use this collection of smooth matrix estimates to compute pairwise distances on the rows and columns based on a new multi-scale metric that implicitly introduces a coupling between the rows and the columns. Finally, we construct row and column representations from these multi-scale metrics. We demonstrate that our approach outperforms competing methods in both data visualization and clustering.'
volume: 97
URL: https://proceedings.mlr.press/v97/mishne19a.html
PDF: http://proceedings.mlr.press/v97/mishne19a/mishne19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mishne19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Gal
family: Mishne
- given: Eric
family: Chi
- given: Ronald
family: Coifman
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4605-4614
id: mishne19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4605
lastpage: 4614
published: 2019-05-24 00:00:00 +0000
- title: 'Agnostic Federated Learning'
abstract: 'A key learning scenario in large-scale applications is that of federated learning, where a centralized model is trained based on data originating from a large number of clients. We argue that, with the existing training and inference, federated models can be biased towards different clients. Instead, we propose a new framework of agnostic federated learning, where the centralized model is optimized for any target distribution formed by a mixture of the client distributions. We further show that this framework naturally yields a notion of fairness. We present data-dependent Rademacher complexity guarantees for learning with this objective, which guide the definition of an algorithm for agnostic federated learning. We also give a fast stochastic optimization algorithm for solving the corresponding optimization problem, for which we prove convergence bounds, assuming a convex loss function and a convex hypothesis set. We further empirically demonstrate the benefits of our approach in several datasets. Beyond federated learning, our framework and algorithm can be of interest to other learning scenarios such as cloud computing, domain adaptation, drifting, and other contexts where the training and test distributions do not coincide.'
volume: 97
URL: https://proceedings.mlr.press/v97/mohri19a.html
PDF: http://proceedings.mlr.press/v97/mohri19a/mohri19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mohri19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Mehryar
family: Mohri
- given: Gary
family: Sivek
- given: Ananda Theertha
family: Suresh
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4615-4625
id: mohri19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4615
lastpage: 4625
published: 2019-05-24 00:00:00 +0000
- title: 'Flat Metric Minimization with Applications in Generative Modeling'
abstract: 'We take the novel perspective to view data not as a probability distribution but rather as a current. Primarily studied in the field of geometric measure theory, k-currents are continuous linear functionals acting on compactly supported smooth differential forms and can be understood as a generalized notion of oriented k-dimensional manifold. By moving from distributions (which are 0-currents) to k-currents, we can explicitly orient the data by attaching a k-dimensional tangent plane to each sample point. Based on the flat metric which is a fundamental distance between currents, we derive FlatGAN, a formulation in the spirit of generative adversarial networks but generalized to k-currents. In our theoretical contribution we prove that the flat metric between a parametrized current and a reference current is Lipschitz continuous in the parameters. In experiments, we show that the proposed shift to k>0 leads to interpretable and disentangled latent representations which behave equivariantly to the specified oriented tangent planes.'
volume: 97
URL: https://proceedings.mlr.press/v97/mollenhoff19a.html
PDF: http://proceedings.mlr.press/v97/mollenhoff19a/mollenhoff19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mollenhoff19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Thomas
family: Möllenhoff
- given: Daniel
family: Cremers
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4626-4635
id: mollenhoff19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4626
lastpage: 4635
published: 2019-05-24 00:00:00 +0000
- title: 'Parsimonious Black-Box Adversarial Attacks via Efficient Combinatorial Optimization'
abstract: 'Solving for adversarial examples with projected gradient descent has been demonstrated to be highly effective in fooling the neural network based classifiers. However, in the black-box setting, the attacker is limited only to the query access to the network and solving for a successful adversarial example becomes much more difficult. To this end, recent methods aim at estimating the true gradient signal based on the input queries but at the cost of excessive queries. We propose an efficient discrete surrogate to the optimization problem which does not require estimating the gradient and consequently becomes free of the first order update hyperparameters to tune. Our experiments on Cifar-10 and ImageNet show the state of the art black-box attack performance with significant reduction in the required queries compared to a number of recently proposed methods. The source code is available at https://github.com/snu-mllab/parsimonious-blackbox-attack.'
volume: 97
URL: https://proceedings.mlr.press/v97/moon19a.html
PDF: http://proceedings.mlr.press/v97/moon19a/moon19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-moon19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Seungyong
family: Moon
- given: Gaon
family: An
- given: Hyun Oh
family: Song
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4636-4645
id: moon19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4636
lastpage: 4645
published: 2019-05-24 00:00:00 +0000
- title: 'Parameter efficient training of deep convolutional neural networks by dynamic sparse reparameterization'
abstract: 'Modern deep neural networks are typically highly overparameterized. Pruning techniques are able to remove a significant fraction of network parameters with little loss in accuracy. Recently, techniques based on dynamic reallocation of non-zero parameters have emerged, allowing direct training of sparse networks without having to pre-train a large dense model. Here we present a novel dynamic sparse reparameterization method that addresses the limitations of previous techniques such as high computational cost and the need for manual configuration of the number of free parameters allocated to each layer. We evaluate the performance of dynamic reallocation methods in training deep convolutional networks and show that our method outperforms previous static and dynamic reparameterization methods, yielding the best accuracy for a fixed parameter budget, on par with accuracies obtained by iteratively pruning a pre-trained dense model. We further investigated the mechanisms underlying the superior generalization performance of the resultant sparse networks. We found that neither the structure, nor the initialization of the non-zero parameters were sufficient to explain the superior performance. Rather, effective learning crucially depended on the continuous exploration of the sparse network structure space during training. Our work suggests that exploring structural degrees of freedom during training is more effective than adding extra parameters to the network.'
volume: 97
URL: https://proceedings.mlr.press/v97/mostafa19a.html
PDF: http://proceedings.mlr.press/v97/mostafa19a/mostafa19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-mostafa19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Hesham
family: Mostafa
- given: Xin
family: Wang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4646-4655
id: mostafa19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4646
lastpage: 4655
published: 2019-05-24 00:00:00 +0000
- title: 'A Dynamical Systems Perspective on Nesterov Acceleration'
abstract: 'We present a dynamical system framework for understanding Nesterov’s accelerated gradient method. In contrast to earlier work, our derivation does not rely on a vanishing step size argument. We show that Nesterov acceleration arises from discretizing an ordinary differential equation with a semi-implicit Euler integration scheme. We analyze both the underlying differential equation as well as the discretization to obtain insights into the phenomenon of acceleration. The analysis suggests that a curvature-dependent damping term lies at the heart of the phenomenon. We further establish connections between the discretized and the continuous-time dynamics.'
volume: 97
URL: https://proceedings.mlr.press/v97/muehlebach19a.html
PDF: http://proceedings.mlr.press/v97/muehlebach19a/muehlebach19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-muehlebach19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Michael
family: Muehlebach
- given: Michael
family: Jordan
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4656-4662
id: muehlebach19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4656
lastpage: 4662
published: 2019-05-24 00:00:00 +0000
- title: 'Relational Pooling for Graph Representations'
abstract: 'This work generalizes graph neural networks (GNNs) beyond those based on the Weisfeiler-Lehman (WL) algorithm, graph Laplacians, and diffusions. Our approach, denoted Relational Pooling (RP), draws from the theory of finite partial exchangeability to provide a framework with maximal representation power for graphs. RP can work with existing graph representation models and, somewhat counterintuitively, can make them even more powerful than the original WL isomorphism test. Additionally, RP allows architectures like Recurrent Neural Networks and Convolutional Neural Networks to be used in a theoretically sound approach for graph classification. We demonstrate improved performance of RP-based graph representations over state-of-the-art methods on a number of tasks.'
volume: 97
URL: https://proceedings.mlr.press/v97/murphy19a.html
PDF: http://proceedings.mlr.press/v97/murphy19a/murphy19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-murphy19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ryan
family: Murphy
- given: Balasubramaniam
family: Srinivasan
- given: Vinayak
family: Rao
- given: Bruno
family: Ribeiro
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4663-4673
id: murphy19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4663
lastpage: 4673
published: 2019-05-24 00:00:00 +0000
- title: 'Learning Optimal Fair Policies'
abstract: 'Systematic discriminatory biases present in our society influence the way data is collected and stored, the way variables are defined, and the way scientific findings are put into practice as policy. Automated decision procedures and learning algorithms applied to such data may serve to perpetuate existing injustice or unfairness in our society. In this paper, we consider how to make optimal but fair decisions, which “break the cycle of injustice” by correcting for the unfair dependence of both decisions and outcomes on sensitive features (e.g., variables that correspond to gender, race, disability, or other protected attributes). We use methods from causal inference and constrained optimization to learn optimal policies in a way that addresses multiple potential biases which afflict data analysis in sensitive contexts, extending the approach of Nabi & Shpitser (2018). Our proposal comes equipped with the theoretical guarantee that the chosen fair policy will induce a joint distribution for new instances that satisfies given fairness constraints. We illustrate our approach with both synthetic data and real criminal justice data.'
volume: 97
URL: https://proceedings.mlr.press/v97/nabi19a.html
PDF: http://proceedings.mlr.press/v97/nabi19a/nabi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nabi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Razieh
family: Nabi
- given: Daniel
family: Malinsky
- given: Ilya
family: Shpitser
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4674-4682
id: nabi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4674
lastpage: 4682
published: 2019-05-24 00:00:00 +0000
- title: 'Lexicographic and Depth-Sensitive Margins in Homogeneous and Non-Homogeneous Deep Models'
abstract: 'With an eye toward understanding complexity control in deep learning, we study how infinitesimal regularization or gradient descent optimization lead to margin maximizing solutions in both homogeneous and non homogeneous models, extending previous work that focused on infinitesimal regularization only in homogeneous models. To this end we study the limit of loss minimization with a diverging norm constraint (the “constrained path”), relate it to the limit of a “margin path” and characterize the resulting solution. For non-homogeneous ensemble models, which output is a sum of homogeneous sub-models, we show that this solution discards the shallowest sub-models if they are unnecessary. For homogeneous models, we show convergence to a “lexicographic max-margin solution”, and provide conditions under which max-margin solutions are also attained as the limit of unconstrained gradient descent.'
volume: 97
URL: https://proceedings.mlr.press/v97/nacson19a.html
PDF: http://proceedings.mlr.press/v97/nacson19a/nacson19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nacson19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Mor Shpigel
family: Nacson
- given: Suriya
family: Gunasekar
- given: Jason
family: Lee
- given: Nathan
family: Srebro
- given: Daniel
family: Soudry
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4683-4692
id: nacson19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4683
lastpage: 4692
published: 2019-05-24 00:00:00 +0000
- title: 'A Wrapped Normal Distribution on Hyperbolic Space for Gradient-Based Learning'
abstract: 'Hyperbolic space is a geometry that is known to be well-suited for representation learning of data with an underlying hierarchical structure. In this paper, we present a novel hyperbolic distribution called hyperbolic wrapped distribution, a wrapped normal distribution on hyperbolic space whose density can be evaluated analytically and differentiated with respect to the parameters. Our distribution enables the gradient-based learning of the probabilistic models on hyperbolic space that could never have been considered before. Also, we can sample from this hyperbolic probability distribution without resorting to auxiliary means like rejection sampling. As applications of our distribution, we develop a hyperbolic-analog of variational autoencoder and a method of probabilistic word embedding on hyperbolic space. We demonstrate the efficacy of our distribution on various datasets including MNIST, Atari 2600 Breakout, and WordNet.'
volume: 97
URL: https://proceedings.mlr.press/v97/nagano19a.html
PDF: http://proceedings.mlr.press/v97/nagano19a/nagano19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nagano19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yoshihiro
family: Nagano
- given: Shoichiro
family: Yamaguchi
- given: Yasuhiro
family: Fujita
- given: Masanori
family: Koyama
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4693-4702
id: nagano19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4693
lastpage: 4702
published: 2019-05-24 00:00:00 +0000
- title: 'SGD without Replacement: Sharper Rates for General Smooth Convex Functions'
abstract: 'We study stochastic gradient descent *without replacement* (SGDo) for smooth convex functions. SGDo is widely observed to converge faster than true SGD where each sample is drawn independently *with replacement* (Bottou,2009) and hence, is more popular in practice. But it’s convergence properties are not well understood as sampling without replacement leads to coupling between iterates and gradients. By using method of exchangeable pairs to bound Wasserstein distance, we provide the first non-asymptotic results for SGDo when applied to *general smooth, strongly-convex* functions. In particular, we show that SGDo converges at a rate of $O(1/K^2)$ while SGD is known to converge at $O(1/K)$ rate, where $K$ denotes the number of passes over data and is required to be *large enough*. Existing results for SGDo in this setting require additional *Hessian Lipschitz assumption* (Gurbuzbalaban et al, 2015; HaoChen and Sra 2018). For *small* $K$, we show SGDo can achieve same convergence rate as SGD for *general smooth strongly-convex* functions. Existing results in this setting require $K=1$ and hold only for generalized linear models (Shamir,2016). In addition, by careful analysis of the coupling, for both large and small $K$, we obtain better dependence on problem dependent parameters like condition number.'
volume: 97
URL: https://proceedings.mlr.press/v97/nagaraj19a.html
PDF: http://proceedings.mlr.press/v97/nagaraj19a/nagaraj19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nagaraj19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Dheeraj
family: Nagaraj
- given: Prateek
family: Jain
- given: Praneeth
family: Netrapalli
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4703-4711
id: nagaraj19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4703
lastpage: 4711
published: 2019-05-24 00:00:00 +0000
- title: 'Dropout as a Structured Shrinkage Prior'
abstract: 'Dropout regularization of deep neural networks has been a mysterious yet effective tool to prevent overfitting. Explanations for its success range from the prevention of "co-adapted" weights to it being a form of cheap Bayesian inference. We propose a novel framework for understanding multiplicative noise in neural networks, considering continuous distributions as well as Bernoulli noise (i.e. dropout). We show that multiplicative noise induces structured shrinkage priors on a network’s weights. We derive the equivalence through reparametrization properties of scale mixtures and without invoking any approximations. Given the equivalence, we then show that dropout’s Monte Carlo training objective approximates marginal MAP estimation. We leverage these insights to propose a novel shrinkage framework for resnets, terming the prior ’automatic depth determination’ as it is the natural analog of automatic relevance determination for network depth. Lastly, we investigate two inference strategies that improve upon the aforementioned MAP approximation in regression benchmarks.'
volume: 97
URL: https://proceedings.mlr.press/v97/nalisnick19a.html
PDF: http://proceedings.mlr.press/v97/nalisnick19a/nalisnick19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nalisnick19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Eric
family: Nalisnick
- given: Jose Miguel
family: Hernandez-Lobato
- given: Padhraic
family: Smyth
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4712-4722
id: nalisnick19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4712
lastpage: 4722
published: 2019-05-24 00:00:00 +0000
- title: 'Hybrid Models with Deep and Invertible Features'
abstract: 'We propose a neural hybrid model consisting of a linear model defined on a set of features computed by a deep, invertible transformation (i.e. a normalizing flow). An attractive property of our model is that both p(features), the density of the features, and p(targets|features), the predictive distribution, can be computed exactly in a single feed-forward pass. We show that our hybrid model, despite the invertibility constraints, achieves similar accuracy to purely predictive models. Yet the generative component remains a good model of the input features despite the hybrid optimization objective. This offers additional capabilities such as detection of out-of-distribution inputs and enabling semi-supervised learning. The availability of the exact joint density p(targets, features) also allows us to compute many quantities readily, making our hybrid model a useful building block for downstream applications of probabilistic deep learning.'
volume: 97
URL: https://proceedings.mlr.press/v97/nalisnick19b.html
PDF: http://proceedings.mlr.press/v97/nalisnick19b/nalisnick19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nalisnick19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Eric
family: Nalisnick
- given: Akihiro
family: Matsukawa
- given: Yee Whye
family: Teh
- given: Dilan
family: Gorur
- given: Balaji
family: Lakshminarayanan
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4723-4732
id: nalisnick19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4723
lastpage: 4732
published: 2019-05-24 00:00:00 +0000
- title: 'Learning Context-dependent Label Permutations for Multi-label Classification'
abstract: 'A key problem in multi-label classification is to utilize dependencies among the labels. Chaining classifiers are a simple technique for addressing this problem but current algorithms all assume a fixed, static label ordering. In this work, we propose a multi-label classification approach which allows to choose a dynamic, context-dependent label ordering. Our proposed approach consists of two sub-components: a simple EM-like algorithm which bootstraps the learned model, and a more elaborate approach based on reinforcement learning. Our experiments on three public multi-label classification benchmarks show that our proposed dynamic label ordering approach based on reinforcement learning outperforms recurrent neural networks with fixed label ordering across both bipartition and ranking measures on all the three datasets. As a result, we obtain a powerful sequence prediction-based algorithm for multi-label classification, which is able to efficiently and explicitly exploit label dependencies.'
volume: 97
URL: https://proceedings.mlr.press/v97/nam19a.html
PDF: http://proceedings.mlr.press/v97/nam19a/nam19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nam19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jinseok
family: Nam
- given: Young-Bum
family: Kim
- given: Eneldo Loza
family: Mencia
- given: Sunghyun
family: Park
- given: Ruhi
family: Sarikaya
- given: Johannes
family: Fürnkranz
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4733-4742
id: nam19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4733
lastpage: 4742
published: 2019-05-24 00:00:00 +0000
- title: 'Zero-Shot Knowledge Distillation in Deep Networks'
abstract: 'Knowledge distillation deals with the problem of training a smaller model (*Student*) from a high capacity source model (*Teacher*) so as to retain most of its performance. Existing approaches use either the training data or meta-data extracted from it in order to train the *Student*. However, accessing the dataset on which the *Teacher* has been trained may not always be feasible if the dataset is very large or it poses privacy or safety concerns (e.g., bio-metric or medical data). Hence, in this paper, we propose a novel data-free method to train the *Student* from the *Teacher*. Without even using any meta-data, we synthesize the *Data Impressions* from the complex *Teacher* model and utilize these as surrogates for the original training data samples to transfer its learning to *Student* via knowledge distillation. We, therefore, dub our method “Zero-Shot Knowledge Distillation" and demonstrate that our framework results in competitive generalization performance as achieved by distillation using the actual training data samples on multiple benchmark datasets.'
volume: 97
URL: https://proceedings.mlr.press/v97/nayak19a.html
PDF: http://proceedings.mlr.press/v97/nayak19a/nayak19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nayak19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Gaurav Kumar
family: Nayak
- given: Konda Reddy
family: Mopuri
- given: Vaisakh
family: Shaj
- given: Venkatesh Babu
family: Radhakrishnan
- given: Anirban
family: Chakraborty
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4743-4751
id: nayak19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4743
lastpage: 4751
published: 2019-05-24 00:00:00 +0000
- title: 'A Framework for Bayesian Optimization in Embedded Subspaces'
abstract: 'We present a theoretically founded approach for high-dimensional Bayesian optimization based on low-dimensional subspace embeddings. We prove that the error in the Gaussian process model is bounded tightly when going from the original high-dimensional search domain to the low-dimensional embedding. This implies that the optimization process in the low-dimensional embedding proceeds essentially as if it were run directly on an unknown active subspace of low dimensionality. The argument applies to a large class of algorithms and GP models, including non-stationary kernels. Moreover, we provide an efficient implementation based on hashing and demonstrate empirically that this subspace embedding achieves considerably better results than the previously proposed methods for high-dimensional BO based on Gaussian matrix projections and structure-learning.'
volume: 97
URL: https://proceedings.mlr.press/v97/nayebi19a.html
PDF: http://proceedings.mlr.press/v97/nayebi19a/nayebi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nayebi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Amin
family: Nayebi
- given: Alexander
family: Munteanu
- given: Matthias
family: Poloczek
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4752-4761
id: nayebi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4752
lastpage: 4761
published: 2019-05-24 00:00:00 +0000
- title: 'Phaseless PCA: Low-Rank Matrix Recovery from Column-wise Phaseless Measurements'
abstract: 'This work proposes the first set of simple, practically useful, and provable algorithms for two inter-related problems. (i) The first is low-rank matrix recovery from magnitude-only (phaseless) linear projections of each of its columns. This finds important applications in phaseless dynamic imaging, e.g., Fourier Ptychographic imaging of live biological specimens. Our guarantee shows that, in the regime of small ranks, the sample complexity required is only a little larger than the order-optimal one, and much smaller than what standard (unstructured) phase retrieval methods need. %Moreover our algorithm is fast and memory-efficient if only the minimum required number of measurements is used (ii) The second problem we study is a dynamic extension of the above: it allows the low-dimensional subspace from which each image/signal (each column of the low-rank matrix) is generated to change with time. We introduce a simple algorithm that is provably correct as long as the subspace changes are piecewise constant.'
volume: 97
URL: https://proceedings.mlr.press/v97/nayer19a.html
PDF: http://proceedings.mlr.press/v97/nayer19a/nayer19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nayer19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Seyedehsara
family: Nayer
- given: Praneeth
family: Narayanamurthy
- given: Namrata
family: Vaswani
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4762-4770
id: nayer19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4762
lastpage: 4770
published: 2019-05-24 00:00:00 +0000
- title: 'Safe Grid Search with Optimal Complexity'
abstract: 'Popular machine learning estimators involve regularization parameters that can be challenging to tune, and standard strategies rely on grid search for this task. In this paper, we revisit the techniques of approximating the regularization path up to predefined tolerance $\epsilon$ in a unified framework and show that its complexity is $O(1/\sqrt[d]{\epsilon})$ for uniformly convex loss of order $d \geq 2$ and $O(1/\sqrt{\epsilon})$ for Generalized Self-Concordant functions. This framework encompasses least-squares but also logistic regression, a case that as far as we know was not handled as precisely in previous works. We leverage our technique to provide refined bounds on the validation error as well as a practical algorithm for hyperparameter tuning. The latter has global convergence guarantee when targeting a prescribed accuracy on the validation set. Last but not least, our approach helps relieving the practitioner from the (often neglected) task of selecting a stopping criterion when optimizing over the training set: our method automatically calibrates this criterion based on the targeted accuracy on the validation set.'
volume: 97
URL: https://proceedings.mlr.press/v97/ndiaye19a.html
PDF: http://proceedings.mlr.press/v97/ndiaye19a/ndiaye19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ndiaye19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Eugene
family: Ndiaye
- given: Tam
family: Le
- given: Olivier
family: Fercoq
- given: Joseph
family: Salmon
- given: Ichiro
family: Takeuchi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4771-4780
id: ndiaye19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4771
lastpage: 4780
published: 2019-05-24 00:00:00 +0000
- title: 'Learning to bid in revenue-maximizing auctions'
abstract: 'We consider the problem of the optimization of bidding strategies in prior-dependent revenue-maximizing auctions, when the seller fixes the reserve prices based on the bid distributions. Our study is done in the setting where one bidder is strategic. Using a variational approach, we study the complexity of the original objective and we introduce a relaxation of the objective functional in order to use gradient descent methods. Our approach is simple, general and can be applied to various value distributions and revenue-maximizing mechanisms. The new strategies we derive yield massive uplifts compared to the traditional truthfully bidding strategy.'
volume: 97
URL: https://proceedings.mlr.press/v97/nedelec19a.html
PDF: http://proceedings.mlr.press/v97/nedelec19a/nedelec19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nedelec19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Thomas
family: Nedelec
- given: Noureddine El
family: Karoui
- given: Vianney
family: Perchet
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4781-4789
id: nedelec19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4781
lastpage: 4789
published: 2019-05-24 00:00:00 +0000
- title: 'On Connected Sublevel Sets in Deep Learning'
abstract: 'This paper shows that every sublevel set of the loss function of a class of deep over-parameterized neural nets with piecewise linear activation functions is connected and unbounded. This implies that the loss has no bad local valleys and all of its global minima are connected within a unique and potentially very large global valley.'
volume: 97
URL: https://proceedings.mlr.press/v97/nguyen19a.html
PDF: http://proceedings.mlr.press/v97/nguyen19a/nguyen19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nguyen19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Quynh
family: Nguyen
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4790-4799
id: nguyen19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4790
lastpage: 4799
published: 2019-05-24 00:00:00 +0000
- title: 'Anomaly Detection With Multiple-Hypotheses Predictions'
abstract: 'In one-class-learning tasks, only the normal case (foreground) can be modeled with data, whereas the variation of all possible anomalies is too erratic to be described by samples. Thus, due to the lack of representative data, the wide-spread discriminative approaches cannot cover such learning tasks, and rather generative models,which attempt to learn the input density of the foreground, are used. However, generative models suffer from a large input dimensionality (as in images) and are typically inefficient learners.We propose to learn the data distribution of the foreground more efficiently with a multi-hypotheses autoencoder. Moreover, the model is criticized by a discriminator, which prevents artificial data modes not supported by data, and which enforces diversity across hypotheses. Our multiple-hypotheses-based anomaly detection framework allows the reliable identification of out-of-distribution samples. For anomaly detection on CIFAR-10, it yields up to 3.9% points improvement over previously reported results. On a real anomaly detection task, the approach reduces the error of the baseline models from 6.8% to 1.5%.'
volume: 97
URL: https://proceedings.mlr.press/v97/nguyen19b.html
PDF: http://proceedings.mlr.press/v97/nguyen19b/nguyen19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nguyen19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Duc Tam
family: Nguyen
- given: Zhongyu
family: Lou
- given: Michael
family: Klar
- given: Thomas
family: Brox
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4800-4809
id: nguyen19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4800
lastpage: 4809
published: 2019-05-24 00:00:00 +0000
- title: 'Non-Asymptotic Analysis of Fractional Langevin Monte Carlo for Non-Convex Optimization'
abstract: 'Recent studies on diffusion-based sampling methods have shown that Langevin Monte Carlo (LMC) algorithms can be beneficial for non-convex optimization, and rigorous theoretical guarantees have been proven for both asymptotic and finite-time regimes. Algorithmically, LMC-based algorithms resemble the well-known gradient descent (GD) algorithm, where the GD recursion is perturbed by an additive Gaussian noise whose variance has a particular form. Fractional Langevin Monte Carlo (FLMC) is a recently proposed extension of LMC, where the Gaussian noise is replaced by a heavy-tailed $\alpha$-stable noise. As opposed to its Gaussian counterpart, these heavy-tailed perturbations can incur large jumps and it has been empirically demonstrated that the choice of $\alpha$-stable noise can provide several advantages in modern machine learning problems, both in optimization and sampling contexts. However, as opposed to LMC, only asymptotic convergence properties of FLMC have been yet established. In this study, we analyze the non-asymptotic behavior of FLMC for non-convex optimization and prove finite-time bounds for its expected suboptimality. Our results show that the weak-error of FLMC increases faster than LMC, which suggests using smaller step-sizes in FLMC. We finally extend our results to the case where the exact gradients are replaced by stochastic gradients and show that similar results hold in this setting as well.'
volume: 97
URL: https://proceedings.mlr.press/v97/nguyen19c.html
PDF: http://proceedings.mlr.press/v97/nguyen19c/nguyen19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nguyen19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Than Huy
family: Nguyen
- given: Umut
family: Simsekli
- given: Gael
family: Richard
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4810-4819
id: nguyen19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4810
lastpage: 4819
published: 2019-05-24 00:00:00 +0000
- title: 'Rotation Invariant Householder Parameterization for Bayesian PCA'
abstract: 'We consider probabilistic PCA and related factor models from a Bayesian perspective. These models are in general not identifiable as the likelihood has a rotational symmetry. This gives rise to complicated posterior distributions with continuous subspaces of equal density and thus hinders efficiency of inference as well as interpretation of obtained parameters. In particular, posterior averages over factor loadings become meaningless and only model predictions are unambiguous. Here, we propose a parameterization based on Householder transformations, which remove the rotational symmetry of the posterior. Furthermore, by relying on results from random matrix theory, we establish the parameter distribution which leaves the model unchanged compared to the original rotationally symmetric formulation. In particular, we avoid the need to compute the Jacobian determinant of the parameter transformation. This allows us to efficiently implement probabilistic PCA in a rotation invariant fashion in any state of the art toolbox. Here, we implemented our model in the probabilistic programming language Stan and illustrate it on several examples.'
volume: 97
URL: https://proceedings.mlr.press/v97/nirwan19a.html
PDF: http://proceedings.mlr.press/v97/nirwan19a/nirwan19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nirwan19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Rajbir
family: Nirwan
- given: Nils
family: Bertschinger
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4820-4828
id: nirwan19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4820
lastpage: 4828
published: 2019-05-24 00:00:00 +0000
- title: 'Lossless or Quantized Boosting with Integer Arithmetic'
abstract: 'In supervised learning, efficiency often starts with the choice of a good loss: support vector machines popularised Hinge loss, Adaboost popularised the exponential loss, etc. Recent trends in machine learning have highlighted the necessity for training routines to meet tight requirements on communication, bandwidth, energy, operations, encoding, among others. Fitting the often decades-old state of the art training routines into these new constraints does not go without pain and uncertainty or reduction in the original guarantees. Our paper starts with the design of a new strictly proper canonical, twice differentiable loss called the Q-loss. Importantly, its mirror update over (arbitrary) rational inputs uses only integer arithmetics – more precisely, the sole use of $+, -, /, \times, |.|$. We build a learning algorithm which is able, under mild assumptions, to achieve a lossless boosting-compliant training. We give conditions for a quantization of its main memory footprint, weights, to be done while keeping the whole algorithm boosting-compliant. Experiments display that the algorithm can achieve a fast convergence during the early boosting rounds compared to AdaBoost, even with a weight storage that can be 30+ times smaller. Lastly, we show that the Bayes risk of the Q-loss can be used as node splitting criterion for decision trees and guarantees optimal boosting convergence.'
volume: 97
URL: https://proceedings.mlr.press/v97/nock19a.html
PDF: http://proceedings.mlr.press/v97/nock19a/nock19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nock19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Richard
family: Nock
- given: Robert
family: Williamson
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4829-4838
id: nock19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4829
lastpage: 4838
published: 2019-05-24 00:00:00 +0000
- title: 'Training Neural Networks with Local Error Signals'
abstract: 'Supervised training of neural networks for classification is typically performed with a global loss function. The loss function provides a gradient for the output layer, and this gradient is back-propagated to hidden layers to dictate an update direction for the weights. An alternative approach is to train the network with layer-wise loss functions. In this paper we demonstrate, for the first time, that layer-wise training can approach the state-of-the-art on a variety of image datasets. We use single-layer sub-networks and two different supervised loss functions to generate local error signals for the hidden layers, and we show that the combination of these losses help with optimization in the context of local learning. Using local errors could be a step towards more biologically plausible deep learning because the global error does not have to be transported back to hidden layers. A completely backprop free variant outperforms previously reported results among methods aiming for higher biological plausibility.'
volume: 97
URL: https://proceedings.mlr.press/v97/nokland19a.html
PDF: http://proceedings.mlr.press/v97/nokland19a/nokland19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nokland19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Arild
family: Nøkland
- given: Lars Hiller
family: Eidnes
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4839-4850
id: nokland19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4839
lastpage: 4850
published: 2019-05-24 00:00:00 +0000
- title: 'Remember and Forget for Experience Replay'
abstract: 'Experience replay (ER) is a fundamental component of off-policy deep reinforcement learning (RL). ER recalls experiences from past iterations to compute gradient estimates for the current policy, increasing data-efficiency. However, the accuracy of such updates may deteriorate when the policy diverges from past behaviors and can undermine the performance of ER. Many algorithms mitigate this issue by tuning hyper-parameters to slow down policy changes. An alternative is to actively enforce the similarity between policy and the experiences in the replay memory. We introduce Remember and Forget Experience Replay (ReF-ER), a novel method that can enhance RL algorithms with parameterized policies. ReF-ER (1) skips gradients computed from experiences that are too unlikely with the current policy and (2) regulates policy changes within a trust region of the replayed behaviors. We couple ReF-ER with Q-learning, deterministic policy gradient and off-policy gradient methods. We find that ReF-ER consistently improves the performance of continuous-action, off-policy RL on fully observable benchmarks and partially observable flow control problems.'
volume: 97
URL: https://proceedings.mlr.press/v97/novati19a.html
PDF: http://proceedings.mlr.press/v97/novati19a/novati19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-novati19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Guido
family: Novati
- given: Petros
family: Koumoutsakos
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4851-4860
id: novati19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4851
lastpage: 4860
published: 2019-05-24 00:00:00 +0000
- title: 'Learning to Infer Program Sketches'
abstract: 'Our goal is to build systems which write code automatically from the kinds of specifications humans can most easily provide, such as examples and natural language instruction. The key idea of this work is that a flexible combination of pattern recognition and explicit reasoning can be used to solve these complex programming problems. We propose a method for dynamically integrating these types of information. Our novel intermediate representation and training algorithm allow a program synthesis system to learn, without direct supervision, when to rely on pattern recognition and when to perform symbolic search. Our model matches the memorization and generalization performance of neural synthesis and symbolic search, respectively, and achieves state-of-the-art performance on a dataset of simple English description-to-code programming problems.'
volume: 97
URL: https://proceedings.mlr.press/v97/nye19a.html
PDF: http://proceedings.mlr.press/v97/nye19a/nye19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-nye19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Maxwell
family: Nye
- given: Luke
family: Hewitt
- given: Joshua
family: Tenenbaum
- given: Armando
family: Solar-Lezama
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4861-4870
id: nye19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4861
lastpage: 4870
published: 2019-05-24 00:00:00 +0000
- title: 'Tensor Variable Elimination for Plated Factor Graphs'
abstract: 'A wide class of machine learning algorithms can be reduced to variable elimination on factor graphs. While factor graphs provide a unifying notation for these algorithms, they do not provide a compact way to express repeated structure when compared to plate diagrams for directed graphical models. To exploit efficient tensor algebra in graphs with plates of variables, we generalize undirected factor graphs to plated factor graphs and variable elimination to a tensor variable elimination algorithm that operates directly on plated factor graphs. Moreover, we generalize complexity bounds based on treewidth and characterize the class of plated factor graphs for which inference is tractable. As an application, we integrate tensor variable elimination into the Pyro probabilistic programming language to enable exact inference in discrete latent variable models with repeated structure. We validate our methods with experiments on both directed and undirected graphical models, including applications to polyphonic music modeling, animal movement modeling, and latent sentiment analysis.'
volume: 97
URL: https://proceedings.mlr.press/v97/obermeyer19a.html
PDF: http://proceedings.mlr.press/v97/obermeyer19a/obermeyer19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-obermeyer19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Fritz
family: Obermeyer
- given: Eli
family: Bingham
- given: Martin
family: Jankowiak
- given: Neeraj
family: Pradhan
- given: Justin
family: Chiu
- given: Alexander
family: Rush
- given: Noah
family: Goodman
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4871-4880
id: obermeyer19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4871
lastpage: 4880
published: 2019-05-24 00:00:00 +0000
- title: 'Counterfactual Off-Policy Evaluation with Gumbel-Max Structural Causal Models'
abstract: 'We introduce an off-policy evaluation procedure for highlighting episodes where applying a reinforcement learned (RL) policy is likely to have produced a substantially different outcome than the observed policy. In particular, we introduce a class of structural causal models (SCMs) for generating counterfactual trajectories in finite partially observable Markov Decision Processes (POMDPs). We see this as a useful procedure for off-policy “debugging” in high-risk settings (e.g., healthcare); by decomposing the expected difference in reward between the RL and observed policy into specific episodes, we can identify episodes where the counterfactual difference in reward is most dramatic. This in turn can be used to facilitate review of specific episodes by domain experts. We demonstrate the utility of this procedure with a synthetic environment of sepsis management.'
volume: 97
URL: https://proceedings.mlr.press/v97/oberst19a.html
PDF: http://proceedings.mlr.press/v97/oberst19a/oberst19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-oberst19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Michael
family: Oberst
- given: David
family: Sontag
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4881-4890
id: oberst19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4881
lastpage: 4890
published: 2019-05-24 00:00:00 +0000
- title: 'Model Function Based Conditional Gradient Method with Armijo-like Line Search'
abstract: 'The Conditional Gradient Method is generalized to a class of non-smooth non-convex optimization problems with many applications in machine learning. The proposed algorithm iterates by minimizing so-called model functions over the constraint set. Complemented with an Armijo line search procedure, we prove that subsequences converge to a stationary point. The abstract framework of model functions provides great flexibility in the design of concrete algorithms. As special cases, for example, we develop an algorithm for additive composite problems and an algorithm for non-linear composite problems which leads to a Gauss-Newton-type algorithm. Both instances are novel in non-smooth non-convex optimization and come with numerous applications in machine learning. We perform an experiment on a non-linear robust regression problem and discuss the flexibility of the proposed framework in several matrix factorization formulations.'
volume: 97
URL: https://proceedings.mlr.press/v97/ochs19a.html
PDF: http://proceedings.mlr.press/v97/ochs19a/ochs19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ochs19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Peter
family: Ochs
- given: Yura
family: Malitsky
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4891-4900
id: ochs19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4891
lastpage: 4900
published: 2019-05-24 00:00:00 +0000
- title: 'TensorFuzz: Debugging Neural Networks with Coverage-Guided Fuzzing'
abstract: 'Neural networks are difficult to interpret and debug. We introduce testing techniques for neural networks that can discover errors occurring only for rare inputs. Specifically, we develop coverage-guided fuzzing (CGF) methods for neural networks. In CGF, random mutations of inputs are guided by a coverage metric toward the goal of satisfying user-specified constraints. We describe how approximate nearest neighbor (ANN) algorithms can provide this coverage metric for neural networks. We then combine these methods with techniques for property-based testing (PBT). In PBT, one asserts properties that a function should satisfy and the system automatically generates tests exercising those properties. We then apply this system to practical goals including (but not limited to) surfacing broken loss functions in popular GitHub repositories and making performance improvements to TensorFlow. Finally, we release an open source library called TensorFuzz that implements the described techniques.'
volume: 97
URL: https://proceedings.mlr.press/v97/odena19a.html
PDF: http://proceedings.mlr.press/v97/odena19a/odena19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-odena19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Augustus
family: Odena
- given: Catherine
family: Olsson
- given: David
family: Andersen
- given: Ian
family: Goodfellow
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4901-4911
id: odena19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4901
lastpage: 4911
published: 2019-05-24 00:00:00 +0000
- title: 'Scalable Learning in Reproducing Kernel Krein Spaces'
abstract: 'We provide the first mathematically complete derivation of the Nystr{ö}m method for low-rank approximation of indefinite kernels and propose an efficient method for finding an approximate eigendecomposition of such kernel matrices. Building on this result, we devise highly scalable methods for learning in reproducing kernel Krein spaces. The devised approaches provide a principled and theoretically well-founded means to tackle large scale learning problems with indefinite kernels. The main motivation for our work comes from problems with structured representations (e.g., graphs, strings, time-series), where it is relatively easy to devise a pairwise (dis)similarity function based on intuition and/or knowledge of domain experts. Such functions are typically not positive definite and it is often well beyond the expertise of practitioners to verify this condition. The effectiveness of the devised approaches is evaluated empirically using indefinite kernels defined on structured and vectorial data representations.'
volume: 97
URL: https://proceedings.mlr.press/v97/oglic19a.html
PDF: http://proceedings.mlr.press/v97/oglic19a/oglic19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-oglic19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Dino
family: Oglic
- given: Thomas
family: Gärtner
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4912-4921
id: oglic19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4912
lastpage: 4921
published: 2019-05-24 00:00:00 +0000
- title: 'Approximation and non-parametric estimation of ResNet-type convolutional neural networks'
abstract: 'Convolutional neural networks (CNNs) have been shown to achieve optimal approximation and estimation error rates (in minimax sense) in several function classes. However, previous analyzed optimal CNNs are unrealistically wide and difficult to obtain via optimization due to sparse constraints in important function classes, including the Hölder class. We show a ResNet-type CNN can attain the minimax optimal error rates in these classes in more plausible situations – it can be dense, and its width, channel size, and filter size are constant with respect to sample size. The key idea is that we can replicate the learning ability of Fully-connected neural networks (FNNs) by tailored CNNs, as long as the FNNs have *block-sparse* structures. Our theory is general in a sense that we can automatically translate any approximation rate achieved by block-sparse FNNs into that by CNNs. As an application, we derive approximation and estimation error rates of the aformentioned type of CNNs for the Barron and Hölder classes with the same strategy.'
volume: 97
URL: https://proceedings.mlr.press/v97/oono19a.html
PDF: http://proceedings.mlr.press/v97/oono19a/oono19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-oono19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Kenta
family: Oono
- given: Taiji
family: Suzuki
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4922-4931
id: oono19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4922
lastpage: 4931
published: 2019-05-24 00:00:00 +0000
- title: 'Orthogonal Random Forest for Causal Inference'
abstract: 'We propose the orthogonal random forest, an algorithm that combines Neyman-orthogonality to reduce sensitivity with respect to estimation error of nuisance parameters with generalized random forests (Athey et al., 2017)—a flexible non-parametric method for statistical estimation of conditional moment models using random forests. We provide a consistency rate and establish asymptotic normality for our estimator. We show that under mild assumptions on the consistency rate of the nuisance estimator, we can achieve the same error rate as an oracle with a priori knowledge of these nuisance parameters. We show that when the nuisance functions have a locally sparse parametrization, then a local ell_1-penalized regression achieves the required rate. We apply our method to estimate heterogeneous treatment effects from observational data with discrete treatments or continuous treatments, and we show that, unlike prior work, our method provably allows to control for a high-dimensional set of variables under standard sparsity conditions. We also provide a comprehensive empirical evaluation of our algorithm on both synthetic and real data.'
volume: 97
URL: https://proceedings.mlr.press/v97/oprescu19a.html
PDF: http://proceedings.mlr.press/v97/oprescu19a/oprescu19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-oprescu19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Miruna
family: Oprescu
- given: Vasilis
family: Syrgkanis
- given: Zhiwei Steven
family: Wu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4932-4941
id: oprescu19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4932
lastpage: 4941
published: 2019-05-24 00:00:00 +0000
- title: 'Inferring Heterogeneous Causal Effects in Presence of Spatial Confounding'
abstract: 'We address the problem of inferring the causal effect of an exposure on an outcome across space, using observational data. The data is possibly subject to unmeasured confounding variables which, in a standard approach, must be adjusted for by estimating a nuisance function. Here we develop a method that eliminates the nuisance function, while mitigating the resulting errors-in-variables. The result is a robust and accurate inference method for spatially varying heterogeneous causal effects. The properties of the method are demonstrated on synthetic as well as real data from Germany and the US.'
volume: 97
URL: https://proceedings.mlr.press/v97/osama19a.html
PDF: http://proceedings.mlr.press/v97/osama19a/osama19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-osama19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Muhammad
family: Osama
- given: Dave
family: Zachariah
- given: Thomas B.
family: Schön
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4942-4950
id: osama19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4942
lastpage: 4950
published: 2019-05-24 00:00:00 +0000
- title: 'Overparameterized Nonlinear Learning: Gradient Descent Takes the Shortest Path?'
abstract: 'Many modern learning tasks involve fitting nonlinear models which are trained in an overparameterized regime where the parameters of the model exceed the size of the training dataset. Due to this overparameterization, the training loss may have infinitely many global minima and it is critical to understand the properties of the solutions found by first-order optimization schemes such as (stochastic) gradient descent starting from different initializations. In this paper we demonstrate that when the loss has certain properties over a minimally small neighborhood of the initial point, first order methods such as (stochastic) gradient descent have a few intriguing properties: (1) the iterates converge at a geometric rate to a global optima even when the loss is nonconvex, (2) among all global optima of the loss the iterates converge to one with a near minimal distance to the initial point, (3) the iterates take a near direct route from the initial point to this global optimum. As part of our proof technique, we introduce a new potential function which captures the tradeoff between the loss function and the distance to the initial point as the iterations progress. The utility of our general theory is demonstrated for a variety of problem domains spanning low-rank matrix recovery to shallow neural network training.'
volume: 97
URL: https://proceedings.mlr.press/v97/oymak19a.html
PDF: http://proceedings.mlr.press/v97/oymak19a/oymak19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-oymak19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Samet
family: Oymak
- given: Mahdi
family: Soltanolkotabi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4951-4960
id: oymak19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4951
lastpage: 4960
published: 2019-05-24 00:00:00 +0000
- title: 'Multiplicative Weights Updates as a distributed constrained optimization algorithm: Convergence to second-order stationary points almost always'
abstract: 'Non-concave maximization has been the subject of much recent study in the optimization and machine learning communities, specifically in deep learning. Recent papers ([Ge et al. 2015, Lee et al 2017] and references therein) indicate that first order methods work well and avoid saddles points. Results as in [Lee \etal 2017], however, are limited to the *unconstrained* case or for cases where the critical points are in the interior of the feasibility set, which fail to capture some of the most interesting applications. In this paper we focus on *constrained* non-concave maximization. We analyze a variant of a well-established algorithm in machine learning called Multiplicative Weights Update (MWU) for the maximization problem $\max_{\mathbf{x} \in D} P(\mathbf{x})$, where $P$ is non-concave, twice continuously differentiable and $D$ is a product of simplices. We show that MWU converges almost always for small enough stepsizes to critical points that satisfy the second order KKT conditions, by combining techniques from dynamical systems as well as taking advantage of a recent connection between Baum Eagon inequality and MWU [Palaiopanos et al 2017].'
volume: 97
URL: https://proceedings.mlr.press/v97/panageas19a.html
PDF: http://proceedings.mlr.press/v97/panageas19a/panageas19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-panageas19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ioannis
family: Panageas
- given: Georgios
family: Piliouras
- given: Xiao
family: Wang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4961-4969
id: panageas19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4961
lastpage: 4969
published: 2019-05-24 00:00:00 +0000
- title: 'Improving Adversarial Robustness via Promoting Ensemble Diversity'
abstract: 'Though deep neural networks have achieved significant progress on various tasks, often enhanced by model ensemble, existing high-performance models can be vulnerable to adversarial attacks. Many efforts have been devoted to enhancing the robustness of individual networks and then constructing a straightforward ensemble, e.g., by directly averaging the outputs, which ignores the interaction among networks. This paper presents a new method that explores the interaction among individual networks to improve robustness for ensemble models. Technically, we define a new notion of ensemble diversity in the adversarial setting as the diversity among non-maximal predictions of individual members, and present an adaptive diversity promoting (ADP) regularizer to encourage the diversity, which leads to globally better robustness for the ensemble by making adversarial examples difficult to transfer among individual members. Our method is computationally efficient and compatible with the defense methods acting on individual networks. Empirical results on various datasets verify that our method can improve adversarial robustness while maintaining state-of-the-art accuracy on normal examples.'
volume: 97
URL: https://proceedings.mlr.press/v97/pang19a.html
PDF: http://proceedings.mlr.press/v97/pang19a/pang19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-pang19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Tianyu
family: Pang
- given: Kun
family: Xu
- given: Chao
family: Du
- given: Ning
family: Chen
- given: Jun
family: Zhu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4970-4979
id: pang19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4970
lastpage: 4979
published: 2019-05-24 00:00:00 +0000
- title: 'Nonparametric Bayesian Deep Networks with Local Competition'
abstract: 'The aim of this work is to enable inference of deep networks that retain high accuracy for the least possible model complexity, with the latter deduced from the data during inference. To this end, we revisit deep networks that comprise competing linear units, as opposed to nonlinear units that do not entail any form of (local) competition. In this context, our main technical innovation consists in an inferential setup that leverages solid arguments from Bayesian nonparametrics. We infer both the needed set of connections or locally competing sets of units, as well as the required floating-point precision for storing the network parameters. Specifically, we introduce auxiliary discrete latent variables representing which initial network components are actually needed for modeling the data at hand, and perform Bayesian inference over them by imposing appropriate stick-breaking priors. As we experimentally show using benchmark datasets, our approach yields networks with less computational footprint than the state-of-the-art, and with no compromises in predictive accuracy.'
volume: 97
URL: https://proceedings.mlr.press/v97/panousis19a.html
PDF: http://proceedings.mlr.press/v97/panousis19a/panousis19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-panousis19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Konstantinos
family: Panousis
- given: Sotirios
family: Chatzis
- given: Sergios
family: Theodoridis
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4980-4988
id: panousis19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4980
lastpage: 4988
published: 2019-05-24 00:00:00 +0000
- title: 'Optimistic Policy Optimization via Multiple Importance Sampling'
abstract: 'Policy Search (PS) is an effective approach to Reinforcement Learning (RL) for solving control tasks with continuous state-action spaces. In this paper, we address the exploration-exploitation trade-off in PS by proposing an approach based on Optimism in the Face of Uncertainty. We cast the PS problem as a suitable Multi Armed Bandit (MAB) problem, defined over the policy parameter space, and we propose a class of algorithms that effectively exploit the problem structure, by leveraging Multiple Importance Sampling to perform an off-policy estimation of the expected return. We show that the regret of the proposed approach is bounded by $\widetilde{\mathcal{O}}(\sqrt{T})$ for both discrete and continuous parameter spaces. Finally, we evaluate our algorithms on tasks of varying difficulty, comparing them with existing MAB and RL algorithms.'
volume: 97
URL: https://proceedings.mlr.press/v97/papini19a.html
PDF: http://proceedings.mlr.press/v97/papini19a/papini19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-papini19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Matteo
family: Papini
- given: Alberto Maria
family: Metelli
- given: Lorenzo
family: Lupo
- given: Marcello
family: Restelli
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 4989-4999
id: papini19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 4989
lastpage: 4999
published: 2019-05-24 00:00:00 +0000
- title: 'Deep Residual Output Layers for Neural Language Generation'
abstract: 'Many tasks, including language generation, benefit from learning the structure of the output space, particularly when the space of output labels is large and the data is sparse. State-of-the-art neural language models indirectly capture the output space structure in their classifier weights since they lack parameter sharing across output labels. Learning shared output label mappings helps, but existing methods have limited expressivity and are prone to overfitting. In this paper, we investigate the usefulness of more powerful shared mappings for output labels, and propose a deep residual output mapping with dropout between layers to better capture the structure of the output space and avoid overfitting. Evaluations on three language generation tasks show that our output label mapping can match or improve state-of-the-art recurrent and self-attention architectures, and suggest that the classifier does not necessarily need to be high-rank to better model natural language if it is better at capturing the structure of the output space.'
volume: 97
URL: https://proceedings.mlr.press/v97/pappas19a.html
PDF: http://proceedings.mlr.press/v97/pappas19a/pappas19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-pappas19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Nikolaos
family: Pappas
- given: James
family: Henderson
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5000-5011
id: pappas19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5000
lastpage: 5011
published: 2019-05-24 00:00:00 +0000
- title: 'Measurements of Three-Level Hierarchical Structure in the Outliers in the Spectrum of Deepnet Hessians'
abstract: 'We expose a structure in deep classifying neural networks in the derivative of the logits with respect to the parameters of the model, which is used to explain the existence of outliers in the spectrum of the Hessian. Previous works decomposed the Hessian into two components, attributing the outliers to one of them, the so-called Covariance of gradients. We show this term is not a Covariance but a second moment matrix, i.e., it is influenced by means of gradients. These means possess an additive two-way structure that is the source of the outliers in the spectrum. This structure can be used to approximate the principal subspace of the Hessian using certain "averaging" operations, avoiding the need for high-dimensional eigenanalysis. We corroborate this claim across different datasets, architectures and sample sizes.'
volume: 97
URL: https://proceedings.mlr.press/v97/papyan19a.html
PDF: http://proceedings.mlr.press/v97/papyan19a/papyan19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-papyan19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Vardan
family: Papyan
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5012-5021
id: papyan19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5012
lastpage: 5021
published: 2019-05-24 00:00:00 +0000
- title: 'Generalized Majorization-Minimization'
abstract: 'Non-convex optimization is ubiquitous in machine learning. Majorization-Minimization (MM) is a powerful iterative procedure for optimizing non-convex functions that works by optimizing a sequence of bounds on the function. In MM, the bound at each iteration is required to touch the objective function at the optimizer of the previous bound. We show that this touching constraint is unnecessary and overly restrictive. We generalize MM by relaxing this constraint, and propose a new optimization framework, named Generalized Majorization-Minimization (G-MM), that is more flexible. For instance, G-MM can incorporate application-specific biases into the optimization procedure without changing the objective function. We derive G-MM algorithms for several latent variable models and show empirically that they consistently outperform their MM counterparts in optimizing non-convex objectives. In particular, G-MM algorithms appear to be less sensitive to initialization.'
volume: 97
URL: https://proceedings.mlr.press/v97/parizi19a.html
PDF: http://proceedings.mlr.press/v97/parizi19a/parizi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-parizi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sobhan Naderi
family: Parizi
- given: Kun
family: He
- given: Reza
family: Aghajani
- given: Stan
family: Sclaroff
- given: Pedro
family: Felzenszwalb
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5022-5031
id: parizi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5022
lastpage: 5031
published: 2019-05-24 00:00:00 +0000
- title: 'Variational Laplace Autoencoders'
abstract: 'Variational autoencoders employ an amortized inference model to approximate the posterior of latent variables. However, such amortized variational inference faces two challenges: (1) the limited posterior expressiveness of fully-factorized Gaussian assumption and (2) the amortization error of the inference model. We present a novel approach that addresses both challenges. First, we focus on ReLU networks with Gaussian output and illustrate their connection to probabilistic PCA. Building on this observation, we derive an iterative algorithm that finds the mode of the posterior and apply fullcovariance Gaussian posterior approximation centered on the mode. Subsequently, we present a general framework named Variational Laplace Autoencoders (VLAEs) for training deep generative models. Based on the Laplace approximation of the latent variable posterior, VLAEs enhance the expressiveness of the posterior while reducing the amortization error. Empirical results on MNIST, Omniglot, Fashion-MNIST, SVHN and CIFAR10 show that the proposed approach significantly outperforms other recent amortized or iterative methods on the ReLU networks.'
volume: 97
URL: https://proceedings.mlr.press/v97/park19a.html
PDF: http://proceedings.mlr.press/v97/park19a/park19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-park19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yookoon
family: Park
- given: Chris
family: Kim
- given: Gunhee
family: Kim
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5032-5041
id: park19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5032
lastpage: 5041
published: 2019-05-24 00:00:00 +0000
- title: 'The Effect of Network Width on Stochastic Gradient Descent and Generalization: an Empirical Study'
abstract: 'We investigate how the final parameters found by stochastic gradient descent are influenced by over-parameterization. We generate families of models by increasing the number of channels in a base network, and then perform a large hyper-parameter search to study how the test error depends on learning rate, batch size, and network width. We find that the optimal SGD hyper-parameters are determined by a "normalized noise scale," which is a function of the batch size, learning rate, and initialization conditions. In the absence of batch normalization, the optimal normalized noise scale is directly proportional to width. Wider networks, with their higher optimal noise scale, also achieve higher test accuracy. These observations hold for MLPs, ConvNets, and ResNets, and for two different parameterization schemes ("Standard" and "NTK"). We observe a similar trend with batch normalization for ResNets. Surprisingly, since the largest stable learning rate is bounded, the largest batch size consistent with the optimal normalized noise scale decreases as the width increases.'
volume: 97
URL: https://proceedings.mlr.press/v97/park19b.html
PDF: http://proceedings.mlr.press/v97/park19b/park19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-park19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Daniel
family: Park
- given: Jascha
family: Sohl-Dickstein
- given: Quoc
family: Le
- given: Samuel
family: Smith
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5042-5051
id: park19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5042
lastpage: 5051
published: 2019-05-24 00:00:00 +0000
- title: 'Spectral Approximate Inference'
abstract: 'Given a graphical model (GM), computing its partition function is the most essential inference task, but it is computationally intractable in general. To address the issue, iterative approximation algorithms exploring certain local structure/consistency of GM have been investigated as popular choices in practice. However, due to their local/iterative nature, they often output poor approximations or even do not converge, e.g., in low-temperature regimes (hard instances of large parameters). To overcome the limitation, we propose a novel approach utilizing the global spectral feature of GM. Our contribution is two-fold: (a) we first propose a fully polynomial-time approximation scheme (FPTAS) for approximating the partition function of GM associating with a low-rank coupling matrix; (b) for general high-rank GMs, we design a spectral mean-field scheme utilizing (a) as a subroutine, where it approximates a high-rank GM into a product of rank-1 GMs for an efficient approximation of the partition function. The proposed algorithm is more robust in its running time and accuracy than prior methods, i.e., neither suffers from the convergence issue nor depends on hard local structures, as demonstrated in our experiments.'
volume: 97
URL: https://proceedings.mlr.press/v97/park19c.html
PDF: http://proceedings.mlr.press/v97/park19c/park19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-park19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sejun
family: Park
- given: Eunho
family: Yang
- given: Se-Young
family: Yun
- given: Jinwoo
family: Shin
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5052-5061
id: park19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5052
lastpage: 5061
published: 2019-05-24 00:00:00 +0000
- title: 'Self-Supervised Exploration via Disagreement'
abstract: 'Efficient exploration is a long-standing problem in sensorimotor learning. Major advances have been demonstrated in noise-free, non-stochastic domains such as video games and simulation. However, most of these formulations either get stuck in environments with stochastic dynamics or are too inefficient to be scalable to real robotics setups. In this paper, we propose a formulation for exploration inspired by the work in active learning literature. Specifically, we train an ensemble of dynamics models and incentivize the agent to explore such that the disagreement of those ensembles is maximized. This allows the agent to learn skills by exploring in a self-supervised manner without any external reward. Notably, we further leverage the disagreement objective to optimize the agent’s policy in a differentiable manner, without using reinforcement learning, which results in a sample-efficient exploration. We demonstrate the efficacy of this formulation across a variety of benchmark environments including stochastic-Atari, Mujoco and Unity. Finally, we implement our differentiable exploration on a real robot which learns to interact with objects completely from scratch. Project videos and code are at https://pathak22.github.io/exploration-by-disagreement/'
volume: 97
URL: https://proceedings.mlr.press/v97/pathak19a.html
PDF: http://proceedings.mlr.press/v97/pathak19a/pathak19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-pathak19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Deepak
family: Pathak
- given: Dhiraj
family: Gandhi
- given: Abhinav
family: Gupta
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5062-5071
id: pathak19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5062
lastpage: 5071
published: 2019-05-24 00:00:00 +0000
- title: 'Subspace Robust Wasserstein Distances'
abstract: 'Making sense of Wasserstein distances between discrete measures in high-dimensional settings remains a challenge. Recent work has advocated a two-step approach to improve robustness and facilitate the computation of optimal transport, using for instance projections on random real lines, or a preliminary quantization of the measures to reduce the size of their support. We propose in this work a “max-min” robust variant of the Wasserstein distance by considering the maximal possible distance that can be realized between two measures, assuming they can be projected orthogonally on a lower k-dimensional subspace. Alternatively, we show that the corresponding “min-max” OT problem has a tight convex relaxation which can be cast as that of finding an optimal transport plan with a low transportation cost, where the cost is alternatively defined as the sum of the k largest eigenvalues of the second order moment matrix of the displacements (or matchings) corresponding to that plan (the usual OT definition only considers the trace of that matrix). We show that both quantities inherit several favorable properties from the OT geometry. We propose two algorithms to compute the latter formulation using entropic regularization, and illustrate the interest of this approach empirically.'
volume: 97
URL: https://proceedings.mlr.press/v97/paty19a.html
PDF: http://proceedings.mlr.press/v97/paty19a/paty19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-paty19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: François-Pierre
family: Paty
- given: Marco
family: Cuturi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5072-5081
id: paty19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5072
lastpage: 5081
published: 2019-05-24 00:00:00 +0000
- title: 'Fingerprint Policy Optimisation for Robust Reinforcement Learning'
abstract: 'Policy gradient methods ignore the potential value of adjusting environment variables: unobservable state features that are randomly determined by the environment in a physical setting, but are controllable in a simulator. This can lead to slow learning, or convergence to suboptimal policies, if the environment variable has a large impact on the transition dynamics. In this paper, we present fingerprint policy optimisation (FPO), which finds a policy that is optimal in expectation across the distribution of environment variables. The central idea is to use Bayesian optimisation (BO) to actively select the distribution of the environment variable that maximises the improvement generated by each iteration of the policy gradient method. To make this BO practical, we contribute two easy-to-compute low-dimensional fingerprints of the current policy. Our experiments show that FPO can efficiently learn policies that are robust to significant rare events, which are unlikely to be observable under random sampling, but are key to learning good policies.'
volume: 97
URL: https://proceedings.mlr.press/v97/paul19a.html
PDF: http://proceedings.mlr.press/v97/paul19a/paul19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-paul19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Supratik
family: Paul
- given: Michael A.
family: Osborne
- given: Shimon
family: Whiteson
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5082-5091
id: paul19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5082
lastpage: 5091
published: 2019-05-24 00:00:00 +0000
- title: 'COMIC: Multi-view Clustering Without Parameter Selection'
abstract: 'In this paper, we study two challenges in clustering analysis, namely, how to cluster multi-view data and how to perform clustering without parameter selection on cluster size. To this end, we propose a novel objective function to project raw data into one space in which the projection embraces the geometric consistency (GC) and the cluster assignment consistency (CAC). To be specific, the GC aims to learn a connection graph from a projection space wherein the data points are connected if and only if they belong to the same cluster. The CAC aims to minimize the discrepancy of pairwise connection graphs induced from different views based on the view-consensus assumption, *i.e.*, different views could produce the same cluster assignment structure as they are different portraits of the same object. Thanks to the view-consensus derived from the connection graph, our method could achieve promising performance in learning view-specific representation and eliminating the heterogeneous gaps across different views. Furthermore, with the proposed objective, it could learn almost all parameters including the cluster number from data without labor-intensive parameter selection. Extensive experimental results show the promising performance achieved by our method on five datasets comparing with nine state-of-the-art multi-view clustering approaches.'
volume: 97
URL: https://proceedings.mlr.press/v97/peng19a.html
PDF: http://proceedings.mlr.press/v97/peng19a/peng19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-peng19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Xi
family: Peng
- given: Zhenyu
family: Huang
- given: Jiancheng
family: Lv
- given: Hongyuan
family: Zhu
- given: Joey Tianyi
family: Zhou
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5092-5101
id: peng19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5092
lastpage: 5101
published: 2019-05-24 00:00:00 +0000
- title: 'Domain Agnostic Learning with Disentangled Representations'
abstract: 'Unsupervised model transfer has the potential to greatly improve the generalizability of deep models to novel domains. Yet the current literature assumes that the separation of target data into distinct domains is known a priori. In this paper, we propose the task of Domain-Agnostic Learning (DAL): How to transfer knowledge from a labeled source domain to unlabeled data from arbitrary target domains? To tackle this problem, we devise a novel Deep Adversarial Disentangled Autoencoder (DADA) capable of disentangling domain-specific features from class identity. We demonstrate experimentally that when the target domain labels are unknown, DADA leads to state-of-the-art performance on several image classification datasets.'
volume: 97
URL: https://proceedings.mlr.press/v97/peng19b.html
PDF: http://proceedings.mlr.press/v97/peng19b/peng19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-peng19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Xingchao
family: Peng
- given: Zijun
family: Huang
- given: Ximeng
family: Sun
- given: Kate
family: Saenko
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5102-5112
id: peng19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5102
lastpage: 5112
published: 2019-05-24 00:00:00 +0000
- title: 'Collaborative Channel Pruning for Deep Networks'
abstract: 'Deep networks have achieved impressive performance in various domains, but their applications are largely limited by the prohibitive computational overhead. In this paper, we propose a novel algorithm, namely collaborative channel pruning (CCP), to reduce the computational overhead with negligible performance degradation. The joint impact of pruned/preserved channels on the loss function is quantitatively analyzed, and such interchannel dependency is exploited to determine which channels to be pruned. The channel selection problem is then reformulated as a constrained 0-1 quadratic optimization problem, and the Hessian matrix, which is essential in constructing the above optimization, can be efficiently approximated. Empirical evaluation on two benchmark data sets indicates that our proposed CCP algorithm achieves higher classification accuracy with similar computational complexity than other stateof-the-art channel pruning algorithms'
volume: 97
URL: https://proceedings.mlr.press/v97/peng19c.html
PDF: http://proceedings.mlr.press/v97/peng19c/peng19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-peng19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Hanyu
family: Peng
- given: Jiaxiang
family: Wu
- given: Shifeng
family: Chen
- given: Junzhou
family: Huang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5113-5122
id: peng19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5113
lastpage: 5122
published: 2019-05-24 00:00:00 +0000
- title: 'Exploiting structure of uncertainty for efficient matroid semi-bandits'
abstract: 'We improve the efficiency of algorithms for stochastic combinatorial semi-bandits. In most interesting problems, state-of-the-art algorithms take advantage of structural properties of rewards, such as independence. However, while being minimax optimal in terms of regret, these algorithms are intractable. In our paper, we first reduce their implementation to a specific submodular maximization. Then, in case of matroid constraints, we design adapted approximation routines, thereby providing the first efficient algorithms that exploit the reward structure. In particular, we improve the state-of-the-art efficient gap-free regret bound by a factor sqrt(k), where k is the maximum action size. Finally, we show how our improvement translates to more general budgeted combinatorial semi-bandits.'
volume: 97
URL: https://proceedings.mlr.press/v97/perrault19a.html
PDF: http://proceedings.mlr.press/v97/perrault19a/perrault19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-perrault19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Pierre
family: Perrault
- given: Vianney
family: Perchet
- given: Michal
family: Valko
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5123-5132
id: perrault19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5123
lastpage: 5132
published: 2019-05-24 00:00:00 +0000
- title: 'Cognitive model priors for predicting human decisions'
abstract: 'Human decision-making underlies all economic behavior. For the past four decades, human decision-making under uncertainty has continued to be explained by theoretical models based on prospect theory, a framework that was awarded the Nobel Prize in Economic Sciences. However, theoretical models of this kind have developed slowly, and robust, high-precision predictive models of human decisions remain a challenge. While machine learning is a natural candidate for solving these problems, it is currently unclear to what extent it can improve predictions obtained by current theories. We argue that this is mainly due to data scarcity, since noisy human behavior requires massive sample sizes to be accurately captured by off-the-shelf machine learning methods. To solve this problem, what is needed are machine learning models with appropriate inductive biases for capturing human behavior, and larger datasets. We offer two contributions towards this end: first, we construct “cognitive model priors” by pretraining neural networks with synthetic data generated by cognitive models (i.e., theoretical models developed by cognitive psychologists). We find that fine-tuning these networks on small datasets of real human decisions results in unprecedented state-of-the-art improvements on two benchmark datasets. Second, we present the first large-scale dataset for human decision-making, containing over 240,000 human judgments across over 13,000 decision problems. This dataset reveals the circumstances where cognitive model priors are useful, and provides a new standard for benchmarking prediction of human decisions under uncertainty.'
volume: 97
URL: https://proceedings.mlr.press/v97/peterson19a.html
PDF: http://proceedings.mlr.press/v97/peterson19a/peterson19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-peterson19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: David D.
family: Bourgin
- given: Joshua C.
family: Peterson
- given: Daniel
family: Reichman
- given: Stuart J.
family: Russell
- given: Thomas L.
family: Griffiths
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5133-5141
id: peterson19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5133
lastpage: 5141
published: 2019-05-24 00:00:00 +0000
- title: 'Towards Understanding Knowledge Distillation'
abstract: 'Knowledge distillation, i.e., one classifier being trained on the outputs of another classifier, is an empirically very successful technique for knowledge transfer between classifiers. It has even been observed that classifiers learn much faster and more reliably if trained with the outputs of another classifier as soft labels, instead of from ground truth data. So far, however, there is no satisfactory theoretical explanation of this phenomenon. In this work, we provide the first insights into the working mechanisms of distillation by studying the special case of linear and deep linear classifiers. Specifically, we prove a generalization bound that establishes fast convergence of the expected risk of a distillation-trained linear classifier. From the bound and its proof we extract three key factors that determine the success of distillation: * data geometry – geometric properties of the data distribution, in particular class separation, has a direct influence on the convergence speed of the risk; * optimization bias – gradient descent optimization finds a very favorable minimum of the distillation objective; and * strong monotonicity – the expected risk of the student classifier always decreases when the size of the training set grows.'
volume: 97
URL: https://proceedings.mlr.press/v97/phuong19a.html
PDF: http://proceedings.mlr.press/v97/phuong19a/phuong19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-phuong19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Mary
family: Phuong
- given: Christoph
family: Lampert
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5142-5151
id: phuong19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5142
lastpage: 5151
published: 2019-05-24 00:00:00 +0000
- title: 'Temporal Gaussian Mixture Layer for Videos'
abstract: 'We introduce a new convolutional layer named the Temporal Gaussian Mixture (TGM) layer and present how it can be used to efficiently capture longer-term temporal information in continuous activity videos. The TGM layer is a temporal convolutional layer governed by a much smaller set of parameters (e.g., location/variance of Gaussians) that are fully differentiable. We present our fully convolutional video models with multiple TGM layers for activity detection. The extensive experiments on multiple datasets, including Charades and MultiTHUMOS, confirm the effectiveness of TGM layers, significantly outperforming the state-of-the-arts.'
volume: 97
URL: https://proceedings.mlr.press/v97/piergiovanni19a.html
PDF: http://proceedings.mlr.press/v97/piergiovanni19a/piergiovanni19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-piergiovanni19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Aj
family: Piergiovanni
- given: Michael
family: Ryoo
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5152-5161
id: piergiovanni19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5152
lastpage: 5161
published: 2019-05-24 00:00:00 +0000
- title: 'Voronoi Boundary Classification: A High-Dimensional Geometric Approach via Weighted Monte Carlo Integration'
abstract: 'Voronoi cell decompositions provide a classical avenue to classification. Typical approaches however only utilize point-wise cell-membership information by means of nearest neighbor queries and do not utilize further geometric information about Voronoi cells since the computation of Voronoi diagrams is prohibitively expensive in high dimensions. We propose a Monte-Carlo integration based approach that instead computes a weighted integral over the boundaries of Voronoi cells, thus incorporating additional information about the Voronoi cell structure. We demonstrate the scalability of our approach in up to 3072 dimensional spaces and analyze convergence based on the number of Monte Carlo samples and choice of weight functions. Experiments comparing our approach to Nearest Neighbors, SVM and Random Forests indicate that while our approach performs similarly to Random Forests for large data sizes, the algorithm exhibits non-trivial data-dependent performance characteristics for smaller datasets and can be analyzed in terms of a geometric confidence measure, thus adding to the repertoire of geometric approaches to classification while having the benefit of not requiring any model changes or retraining as new training samples or classes are added.'
volume: 97
URL: https://proceedings.mlr.press/v97/polianskii19a.html
PDF: http://proceedings.mlr.press/v97/polianskii19a/polianskii19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-polianskii19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Vladislav
family: Polianskii
- given: Florian T.
family: Pokorny
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5162-5170
id: polianskii19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5162
lastpage: 5170
published: 2019-05-24 00:00:00 +0000
- title: 'On Variational Bounds of Mutual Information'
abstract: 'Estimating and optimizing Mutual Information (MI) is core to many problems in machine learning, but bounding MI in high dimensions is challenging. To establish tractable and scalable objectives, recent work has turned to variational bounds parameterized by neural networks. However, the relationships and tradeoffs between these bounds remains unclear. In this work, we unify these recent developments in a single framework. We find that the existing variational lower bounds degrade when the MI is large, exhibiting either high bias or high variance. To address this problem, we introduce a continuum of lower bounds that encompasses previous bounds and flexibly trades off bias and variance. On high-dimensional, controlled problems, we empirically characterize the bias and variance of the bounds and their gradients and demonstrate the effectiveness of these new bounds for estimation and representation learning.'
volume: 97
URL: https://proceedings.mlr.press/v97/poole19a.html
PDF: http://proceedings.mlr.press/v97/poole19a/poole19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-poole19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ben
family: Poole
- given: Sherjil
family: Ozair
- given: Aaron
family: Van Den Oord
- given: Alex
family: Alemi
- given: George
family: Tucker
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5171-5180
id: poole19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5171
lastpage: 5180
published: 2019-05-24 00:00:00 +0000
- title: 'Hiring Under Uncertainty'
abstract: 'In this paper we introduce the hiring under uncertainty problem to model the questions faced by hiring committees in large enterprises and universities alike. Given a set of $n$ eligible candidates, the decision maker needs to choose the sequence of candidates to make offers so as to hire the $k$ best candidates. However, candidates may choose to reject an offer (for instance, due to a competing offer) and the decision maker has a time limit by which all positions must be filled. Given an estimate of the probabilities of acceptance for each candidate, the hiring under uncertainty problem is to design a strategy of making offers so that the total expected value of all candidates hired by the time limit is maximized. We provide a 2-approximation algorithm for the setting where offers must be made in sequence, an 8-approximation when offers may be made in parallel, and a 10-approximation for the more general stochastic knapsack setting with finite probes.'
volume: 97
URL: https://proceedings.mlr.press/v97/purohit19a.html
PDF: http://proceedings.mlr.press/v97/purohit19a/purohit19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-purohit19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Manish
family: Purohit
- given: Sreenivas
family: Gollapudi
- given: Manish
family: Raghavan
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5181-5189
id: purohit19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5181
lastpage: 5189
published: 2019-05-24 00:00:00 +0000
- title: 'SAGA with Arbitrary Sampling'
abstract: 'We study the problem of minimizing the average of a very large number of smooth functions, which is of key importance in training supervised learning models. One of the most celebrated methods in this context is the SAGA algorithm of Defazio et al. (2014). Despite years of research on the topic, a general-purpose version of SAGA—one that would include arbitrary importance sampling and minibatching schemes—does not exist. We remedy this situation and propose a general and flexible variant of SAGA following the arbitrary sampling paradigm. We perform an iteration complexity analysis of the method, largely possible due to the construction of new stochastic Lyapunov functions. We establish linear convergence rates in the smooth and strongly convex regime, and under certain error bound conditions also in a regime without strong convexity. Our rates match those of the primal-dual method Quartz (Qu et al., 2015) for which an arbitrary sampling analysis is available, which makes a significant step towards closing the gap in our understanding of complexity of primal and dual methods for finite sum problems. Finally, we show through experiments that specific variants of our general SAGA method can perform better in practice than other competing methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/qian19a.html
PDF: http://proceedings.mlr.press/v97/qian19a/qian19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-qian19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Xun
family: Qian
- given: Zheng
family: Qu
- given: Peter
family: Richtárik
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5190-5199
id: qian19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5190
lastpage: 5199
published: 2019-05-24 00:00:00 +0000
- title: 'SGD: General Analysis and Improved Rates'
abstract: 'We propose a general yet simple theorem describing the convergence of SGD under the arbitrary sampling paradigm. Our theorem describes the convergence of an infinite array of variants of SGD, each of which is associated with a specific probability law governing the data selection rule used to form minibatches. This is the first time such an analysis is performed, and most of our variants of SGD were never explicitly considered in the literature before. Our analysis relies on the recently introduced notion of expected smoothness and does not rely on a uniform bound on the variance of the stochastic gradients. By specializing our theorem to different mini-batching strategies, such as sampling with replacement and independent sampling, we derive exact expressions for the stepsize as a function of the mini-batch size. With this we can also determine the mini-batch size that optimizes the total complexity, and show explicitly that as the variance of the stochastic gradient evaluated at the minimum grows, so does the optimal mini-batch size. For zero variance, the optimal mini-batch size is one. Moreover, we prove insightful stepsize-switching rules which describe when one should switch from a constant to a decreasing stepsize regime.'
volume: 97
URL: https://proceedings.mlr.press/v97/qian19b.html
PDF: http://proceedings.mlr.press/v97/qian19b/qian19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-qian19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Robert Mansel
family: Gower
- given: Nicolas
family: Loizou
- given: Xun
family: Qian
- given: Alibek
family: Sailanbayev
- given: Egor
family: Shulgin
- given: Peter
family: Richtárik
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5200-5209
id: qian19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5200
lastpage: 5209
published: 2019-05-24 00:00:00 +0000
- title: 'AutoVC: Zero-Shot Voice Style Transfer with Only Autoencoder Loss'
abstract: 'Despite the progress in voice conversion, many-to-many voice conversion trained on non-parallel data, as well as zero-shot voice conversion, remains under-explored. Deep style transfer algorithms, generative adversarial networks (GAN) in particular, are being applied as new solutions in this field. However, GAN training is very sophisticated and difficult, and there is no strong evidence that its generated speech is of good perceptual quality. In this paper, we propose a new style transfer scheme that involves only an autoencoder with a carefully designed bottleneck. We formally show that this scheme can achieve distribution-matching style transfer by training only on self-reconstruction loss. Based on this scheme, we proposed AutoVC, which achieves state-of-the-art results in many-to-many voice conversion with non-parallel data, and which is the first to perform zero-shot voice conversion.'
volume: 97
URL: https://proceedings.mlr.press/v97/qian19c.html
PDF: http://proceedings.mlr.press/v97/qian19c/qian19c.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-qian19c.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Kaizhi
family: Qian
- given: Yang
family: Zhang
- given: Shiyu
family: Chang
- given: Xuesong
family: Yang
- given: Mark
family: Hasegawa-Johnson
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5210-5219
id: qian19c
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5210
lastpage: 5219
published: 2019-05-24 00:00:00 +0000
- title: 'Fault Tolerance in Iterative-Convergent Machine Learning'
abstract: 'Machine learning (ML) training algorithms often possess an inherent self-correcting behavior due to their iterative- convergent nature. Recent systems exploit this property to achieve adaptability and efficiency in unreliable computing environments by relaxing the consistency of execution and allowing calculation errors to be self-corrected during training. However, the behavior of such systems are only well understood for specific types of calculation errors, such as those caused by staleness, reduced precision, or asynchronicity, and for specific algorithms, such as stochastic gradient descent. In this paper, we develop a general framework to quantify the effects of calculation errors on iterative-convergent algorithms. We then use this framework to derive a worst-case upper bound on the cost of arbitrary perturbations to model parameters during training and to design new strategies for checkpoint-based fault tolerance. Our system, SCAR, can reduce the cost of partial failures by 78%{–}95% when compared with traditional checkpoint-based fault tolerance across a variety of ML models and training algorithms, providing near-optimal performance in recovering from failures.'
volume: 97
URL: https://proceedings.mlr.press/v97/qiao19a.html
PDF: http://proceedings.mlr.press/v97/qiao19a/qiao19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-qiao19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Aurick
family: Qiao
- given: Bryon
family: Aragam
- given: Bingjing
family: Zhang
- given: Eric
family: Xing
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5220-5230
id: qiao19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5220
lastpage: 5230
published: 2019-05-24 00:00:00 +0000
- title: 'Imperceptible, Robust, and Targeted Adversarial Examples for Automatic Speech Recognition'
abstract: 'Adversarial examples are inputs to machine learning models designed by an adversary to cause an incorrect output. So far, adversarial examples have been studied most extensively in the image domain. In this domain, adversarial examples can be constructed by imperceptibly modifying images to cause misclassification, and are practical in the physical world. In contrast, current targeted adversarial examples on speech recognition systems have neither of these properties: humans can easily identify the adversarial perturbations, and they are not effective when played over-the-air. This paper makes progress on both of these fronts. First, we develop effectively imperceptible audio adversarial examples (verified through a human study) by leveraging the psychoacoustic principle of auditory masking, while retaining 100% targeted success rate on arbitrary full-sentence targets. Then, we make progress towards physical-world audio adversarial examples by constructing perturbations which remain effective even after applying highly-realistic simulated environmental distortions.'
volume: 97
URL: https://proceedings.mlr.press/v97/qin19a.html
PDF: http://proceedings.mlr.press/v97/qin19a/qin19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-qin19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yao
family: Qin
- given: Nicholas
family: Carlini
- given: Garrison
family: Cottrell
- given: Ian
family: Goodfellow
- given: Colin
family: Raffel
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5231-5240
id: qin19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5231
lastpage: 5240
published: 2019-05-24 00:00:00 +0000
- title: 'GMNN: Graph Markov Neural Networks'
abstract: 'This paper studies semi-supervised object classification in relational data, which is a fundamental problem in relational data modeling. The problem has been extensively studied in the literature of both statistical relational learning (e.g. relational Markov networks) and graph neural networks (e.g. graph convolutional networks). Statistical relational learning methods can effectively model the dependency of object labels through conditional random fields for collective classification, whereas graph neural networks learn effective object representations for classification through end-to-end training. In this paper, we propose the Graph Markov Neural Network (GMNN) that combines the advantages of both worlds. A GMNN models the joint distribution of object labels with a conditional random field, which can be effectively trained with the variational EM algorithm. In the E-step, one graph neural network learns effective object representations for approximating the posterior distributions of object labels. In the M-step, another graph neural network is used to model the local label dependency. Experiments on object classification, link classification, and unsupervised node representation learning show that GMNN achieves state-of-the-art results.'
volume: 97
URL: https://proceedings.mlr.press/v97/qu19a.html
PDF: http://proceedings.mlr.press/v97/qu19a/qu19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-qu19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Meng
family: Qu
- given: Yoshua
family: Bengio
- given: Jian
family: Tang
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5241-5250
id: qu19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5241
lastpage: 5250
published: 2019-05-24 00:00:00 +0000
- title: 'Nonlinear Distributional Gradient Temporal-Difference Learning'
abstract: 'We devise a distributional variant of gradient temporal-difference (TD) learning. Distributional reinforcement learning has been demonstrated to outperform the regular one in the recent study \citep{bellemare2017distributional}. In the policy evaluation setting, we design two new algorithms called distributional GTD2 and distributional TDC using the Cram{é}r distance on the distributional version of the Bellman error objective function, which inherits advantages of both the nonlinear gradient TD algorithms and the distributional RL approach. In the control setting, we propose the distributional Greedy-GQ using similar derivation. We prove the asymptotic almost-sure convergence of distributional GTD2 and TDC to a local optimal solution for general smooth function approximators, which includes neural networks that have been widely used in recent study to solve the real-life RL problems. In each step, the computational complexity of above three algorithms is linear w.r.t. the number of the parameters of the function approximator, thus can be implemented efficiently for neural networks.'
volume: 97
URL: https://proceedings.mlr.press/v97/qu19b.html
PDF: http://proceedings.mlr.press/v97/qu19b/qu19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-qu19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Chao
family: Qu
- given: Shie
family: Mannor
- given: Huan
family: Xu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5251-5260
id: qu19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5251
lastpage: 5260
published: 2019-05-24 00:00:00 +0000
- title: 'Learning to Collaborate in Markov Decision Processes'
abstract: 'We consider a two-agent MDP framework where agents repeatedly solve a task in a collaborative setting. We study the problem of designing a learning algorithm for the first agent (A1) that facilitates a successful collaboration even in cases when the second agent (A2) is adapting its policy in an unknown way. The key challenge in our setting is that the first agent faces non-stationarity in rewards and transitions because of the adaptive behavior of the second agent. We design novel online learning algorithms for agent A1 whose regret decays as $O(T^{1-\frac{3}{7} \cdot \alpha})$ with $T$ learning episodes provided that the magnitude of agent A2’s policy changes between any two consecutive episodes are upper bounded by $O(T^{-\alpha})$. Here, the parameter $\alpha$ is assumed to be strictly greater than $0$, and we show that this assumption is necessary provided that the *learning parity with noise* problem is computationally hard. We show that sub-linear regret of agent A1 further implies near-optimality of the agents’ joint return for MDPs that manifest the properties of a *smooth* game.'
volume: 97
URL: https://proceedings.mlr.press/v97/radanovic19a.html
PDF: http://proceedings.mlr.press/v97/radanovic19a/radanovic19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-radanovic19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Goran
family: Radanovic
- given: Rati
family: Devidze
- given: David
family: Parkes
- given: Adish
family: Singla
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5261-5270
id: radanovic19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5261
lastpage: 5270
published: 2019-05-24 00:00:00 +0000
- title: 'Meta-Learning Neural Bloom Filters'
abstract: 'There has been a recent trend in training neural networks to replace data structures that have been crafted by hand, with an aim for faster execution, better accuracy, or greater compression. In this setting, a neural data structure is instantiated by training a network over many epochs of its inputs until convergence. In applications where inputs arrive at high throughput, or are ephemeral, training a network from scratch is not practical. This motivates the need for few-shot neural data structures. In this paper we explore the learning of approximate set membership over a set of data in one-shot via meta-learning. We propose a novel memory architecture, the Neural Bloom Filter, which is able to achieve significant compression gains over classical Bloom Filters and existing memory-augmented neural networks.'
volume: 97
URL: https://proceedings.mlr.press/v97/rae19a.html
PDF: http://proceedings.mlr.press/v97/rae19a/rae19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-rae19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Jack
family: Rae
- given: Sergey
family: Bartunov
- given: Timothy
family: Lillicrap
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5271-5280
id: rae19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5271
lastpage: 5280
published: 2019-05-24 00:00:00 +0000
- title: 'Direct Uncertainty Prediction for Medical Second Opinions'
abstract: 'The issue of disagreements amongst human experts is a ubiquitous one in both machine learning and medicine. In medicine, this often corresponds to doctor disagreements on a patient diagnosis. In this work, we show that machine learning models can be successfully trained to give uncertainty scores to data instances that result in high expert disagreements. In particular, they can identify patient cases that would benefit most from a medical second opinion. Our central methodological finding is that Direct Uncertainty Prediction (DUP), training a model to predict an uncertainty score directly from the raw patient features, works better than Uncertainty Via Classification, the two step process of training a classifier and postprocessing the output distribution to give an uncertainty score. We show this both with a theoretical result, and on extensive evaluations on a large scale medical imaging application.'
volume: 97
URL: https://proceedings.mlr.press/v97/raghu19a.html
PDF: http://proceedings.mlr.press/v97/raghu19a/raghu19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-raghu19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Maithra
family: Raghu
- given: Katy
family: Blumer
- given: Rory
family: Sayres
- given: Ziad
family: Obermeyer
- given: Bobby
family: Kleinberg
- given: Sendhil
family: Mullainathan
- given: Jon
family: Kleinberg
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5281-5290
id: raghu19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5281
lastpage: 5290
published: 2019-05-24 00:00:00 +0000
- title: 'Game Theoretic Optimization via Gradient-based Nikaido-Isoda Function'
abstract: 'Computing Nash equilibrium (NE) of multi-player games has witnessed renewed interest due to recent advances in generative adversarial networks. However, computing equilibrium efficiently is challenging. To this end, we introduce the Gradient-based Nikaido-Isoda (GNI) function which serves: (i) as a merit function, vanishing only at the first-order stationary points of each player’s optimization problem, and (ii) provides error bounds to a stationary Nash point. Gradient descent is shown to converge sublinearly to a first-order stationary point of the GNI function. For the particular case of bilinear min-max games and multi-player quadratic games, the GNI function is convex. Hence, the application of gradient descent in this case yields linear convergence to an NE (when one exists). In our numerical experiments, we observe that the GNI formulation always converges to the first-order stationary point of each player’s optimization problem.'
volume: 97
URL: https://proceedings.mlr.press/v97/raghunathan19a.html
PDF: http://proceedings.mlr.press/v97/raghunathan19a/raghunathan19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-raghunathan19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Arvind
family: Raghunathan
- given: Anoop
family: Cherian
- given: Devesh
family: Jha
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5291-5300
id: raghunathan19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5291
lastpage: 5300
published: 2019-05-24 00:00:00 +0000
- title: 'On the Spectral Bias of Neural Networks'
abstract: 'Neural networks are known to be a class of highly expressive functions able to fit even random input-output mappings with 100% accuracy. In this work we present properties of neural networks that complement this aspect of expressivity. By using tools from Fourier analysis, we highlight a learning bias of deep networks towards low frequency functions – i.e. functions that vary globally without local fluctuations – which manifests itself as a frequency-dependent learning speed. Intuitively, this property is in line with the observation that over-parameterized networks prioritize learning simple patterns that generalize across data samples. We also investigate the role of the shape of the data manifold by presenting empirical and theoretical evidence that, somewhat counter-intuitively, learning higher frequencies gets easier with increasing manifold complexity.'
volume: 97
URL: https://proceedings.mlr.press/v97/rahaman19a.html
PDF: http://proceedings.mlr.press/v97/rahaman19a/rahaman19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-rahaman19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Nasim
family: Rahaman
- given: Aristide
family: Baratin
- given: Devansh
family: Arpit
- given: Felix
family: Draxler
- given: Min
family: Lin
- given: Fred
family: Hamprecht
- given: Yoshua
family: Bengio
- given: Aaron
family: Courville
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5301-5310
id: rahaman19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5301
lastpage: 5310
published: 2019-05-24 00:00:00 +0000
- title: 'Look Ma, No Latent Variables: Accurate Cutset Networks via Compilation'
abstract: 'Tractable probabilistic models obviate the need for unreliable approximate inference approaches and as a result often yield accurate query answers in practice. However, most tractable models that achieve state-of-the-art generalization performance (measured using test set likelihood score) use latent variables. Such models admit poly-time marginal (MAR) inference but do not admit poly-time (full) maximum-a-posteriori (MAP) inference. To address this problem, in this paper, we propose a novel approach for inducing cutset networks, a well-known tractable, highly interpretable representation that does not use latent variables and admits linear time MAR as well as MAP inference. Our approach addresses a major limitation of existing techniques that learn cutset networks from data in that their accuracy is quite low as compared to latent variable models such as ensembles of cutset networks and sum-product networks. The key idea in our approach is to construct deep cutset networks by not only learning them from data but also compiling them from a more accurate latent tractable model. We show experimentally that our new approach yields more accurate MAP estimates as compared with existing approaches and significantly improves the test set log-likelihood score of cutset networks bringing them closer in terms of generalization performance to latent variable models.'
volume: 97
URL: https://proceedings.mlr.press/v97/rahman19a.html
PDF: http://proceedings.mlr.press/v97/rahman19a/rahman19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-rahman19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Tahrima
family: Rahman
- given: Shasha
family: Jin
- given: Vibhav
family: Gogate
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5311-5320
id: rahman19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5311
lastpage: 5320
published: 2019-05-24 00:00:00 +0000
- title: 'Does Data Augmentation Lead to Positive Margin?'
abstract: 'Data augmentation (DA) is commonly used during model training, as it significantly improves test error and model robustness. DA artificially expands the training set by applying random noise, rotations, crops, or even adversarial perturbations to the input data. Although DA is widely used, its capacity to provably improve robustness is not fully understood. In this work, we analyze the robustness that DA begets by quantifying the margin that DA enforces on empirical risk minimizers. We first focus on linear separators, and then a class of nonlinear models whose labeling is constant within small convex hulls of data points. We present lower bounds on the number of augmented data points required for non-zero margin, and show that commonly used DA techniques may only introduce significant margin after adding exponentially many points to the data set.'
volume: 97
URL: https://proceedings.mlr.press/v97/rajput19a.html
PDF: http://proceedings.mlr.press/v97/rajput19a/rajput19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-rajput19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Shashank
family: Rajput
- given: Zhili
family: Feng
- given: Zachary
family: Charles
- given: Po-Ling
family: Loh
- given: Dimitris
family: Papailiopoulos
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5321-5330
id: rajput19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5321
lastpage: 5330
published: 2019-05-24 00:00:00 +0000
- title: 'Efficient Off-Policy Meta-Reinforcement Learning via Probabilistic Context Variables'
abstract: 'Deep reinforcement learning algorithms require large amounts of experience to learn an individual task. While meta-reinforcement learning (meta-RL) algorithms can enable agents to learn new skills from small amounts of experience, several major challenges preclude their practicality. Current methods rely heavily on on-policy experience, limiting their sample efficiency. They also lack mechanisms to reason about task uncertainty when adapting to new tasks, limiting their effectiveness on sparse reward problems. In this paper, we address these challenges by developing an off-policy meta-RL algorithm that disentangles task inference and control. In our approach, we perform online probabilistic filtering of latent task variables to infer how to solve a new task from small amounts of experience. This probabilistic interpretation enables posterior sampling for structured and efficient exploration. We demonstrate how to integrate these task variables with off-policy RL algorithms to achieve both meta-training and adaptation efficiency. Our method outperforms prior algorithms in sample efficiency by 20-100X as well as in asymptotic performance on several meta-RL benchmarks.'
volume: 97
URL: https://proceedings.mlr.press/v97/rakelly19a.html
PDF: http://proceedings.mlr.press/v97/rakelly19a/rakelly19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-rakelly19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Kate
family: Rakelly
- given: Aurick
family: Zhou
- given: Chelsea
family: Finn
- given: Sergey
family: Levine
- given: Deirdre
family: Quillen
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5331-5340
id: rakelly19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5331
lastpage: 5340
published: 2019-05-24 00:00:00 +0000
- title: 'Screening rules for Lasso with non-convex Sparse Regularizers'
abstract: 'Leveraging on the convexity of the Lasso problem, screening rules help in accelerating solvers by discarding irrelevant variables, during the optimization process. However, because they provide better theoretical guarantees in identifying relevant variables, several non-convex regularizers for the Lasso have been proposed in the literature. This work is the first that introduces a screening rule strategy into a non-convex Lasso solver. The approach we propose is based on a iterative majorization-minimization (MM) strategy that includes a screening rule in the inner solver and a condition for propagating screened variables between iterations of MM. In addition to improve efficiency of solvers, we also provide guarantees that the inner solver is able to identify the zeros components of its critical point in finite time. Our experimental analysis illustrates the significant computational gain brought by the new screening rule compared to classical coordinate-descent or proximal gradient descent methods.'
volume: 97
URL: https://proceedings.mlr.press/v97/rakotomamonjy19a.html
PDF: http://proceedings.mlr.press/v97/rakotomamonjy19a/rakotomamonjy19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-rakotomamonjy19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alain
family: Rakotomamonjy
- given: Gilles
family: Gasso
- given: Joseph
family: Salmon
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5341-5350
id: rakotomamonjy19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5341
lastpage: 5350
published: 2019-05-24 00:00:00 +0000
- title: 'Topological Data Analysis of Decision Boundaries with Application to Model Selection'
abstract: 'We propose the labeled Cech complex, the plain labeled Vietoris-Rips complex, and the locally scaled labeled Vietoris-Rips complex to perform persistent homology inference of decision boundaries in classification tasks. We provide theoretical conditions and analysis for recovering the homology of a decision boundary from samples. Our main objective is quantification of deep neural network complexity to enable matching of datasets to pre-trained models to facilitate the functioning of AI marketplaces; we report results for experiments using MNIST, FashionMNIST, and CIFAR10.'
volume: 97
URL: https://proceedings.mlr.press/v97/ramamurthy19a.html
PDF: http://proceedings.mlr.press/v97/ramamurthy19a/ramamurthy19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ramamurthy19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Karthikeyan Natesan
family: Ramamurthy
- given: Kush
family: Varshney
- given: Krishnan
family: Mody
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5351-5360
id: ramamurthy19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5351
lastpage: 5360
published: 2019-05-24 00:00:00 +0000
- title: 'HyperGAN: A Generative Model for Diverse, Performant Neural Networks'
abstract: 'We introduce HyperGAN, a generative model that learns to generate all the parameters of a deep neural network. HyperGAN first transforms low dimensional noise into a latent space, which can be sampled from to obtain diverse, performant sets of parameters for a target architecture. We utilize an architecture that bears resemblance to generative adversarial networks, but we evaluate the likelihood of generated samples with a classification loss. This is equivalent to minimizing the KL-divergence between the distribution of generated parameters, and the unknown true parameter distribution. We apply HyperGAN to classification, showing that HyperGAN can learn to generate parameters which solve the MNIST and CIFAR-10 datasets with competitive performance to fully supervised learning, while also generating a rich distribution of effective parameters. We also show that HyperGAN can also provide better uncertainty estimates than standard ensembles. This is evidenced by the ability of HyperGAN-generated ensembles to detect out of distribution data as well as adversarial examples.'
volume: 97
URL: https://proceedings.mlr.press/v97/ratzlaff19a.html
PDF: http://proceedings.mlr.press/v97/ratzlaff19a/ratzlaff19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ratzlaff19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Neale
family: Ratzlaff
- given: Li
family: Fuxin
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5361-5369
id: ratzlaff19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5361
lastpage: 5369
published: 2019-05-24 00:00:00 +0000
- title: 'Efficient On-Device Models using Neural Projections'
abstract: 'Many applications involving visual and language understanding can be effectively solved using deep neural networks. Even though these techniques achieve state-of-the-art results, it is very challenging to apply them on devices with limited memory and computational capacity such as mobile phones, smart watches and IoT. We propose a neural projection approach for training compact on-device neural networks. We introduce "projection" networks that use locality-sensitive projections to generate compact binary representations and learn small neural networks with computationally efficient operations. We design a joint optimization framework where the projection network can be trained from scratch or leverage existing larger neural networks such as feed-forward NNs, CNNs or RNNs. The trained neural projection network can be directly used for inference on device at low memory and computation cost. We demonstrate the effectiveness of this as a general-purpose approach for significantly shrinking memory requirements of different types of neural networks while preserving good accuracy on multiple visual and text classification tasks.'
volume: 97
URL: https://proceedings.mlr.press/v97/ravi19a.html
PDF: http://proceedings.mlr.press/v97/ravi19a/ravi19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ravi19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Sujith
family: Ravi
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5370-5379
id: ravi19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5370
lastpage: 5379
published: 2019-05-24 00:00:00 +0000
- title: 'A Block Coordinate Descent Proximal Method for Simultaneous Filtering and Parameter Estimation'
abstract: 'We propose and analyze a block coordinate descent proximal algorithm (BCD-prox) for simultaneous filtering and parameter estimation of ODE models. As we show on ODE systems with up to d=40 dimensions, as compared to state-of-the-art methods, BCD-prox exhibits increased robustness (to noise, parameter initialization, and hyperparameters), decreased training times, and improved accuracy of both filtered states and estimated parameters. We show how BCD-prox can be used with multistep numerical discretizations, and we establish convergence of BCD-prox under hypotheses that include real systems of interest.'
volume: 97
URL: https://proceedings.mlr.press/v97/raziperchikolaei19a.html
PDF: http://proceedings.mlr.press/v97/raziperchikolaei19a/raziperchikolaei19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-raziperchikolaei19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Ramin
family: Raziperchikolaei
- given: Harish
family: Bhat
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5380-5388
id: raziperchikolaei19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5380
lastpage: 5388
published: 2019-05-24 00:00:00 +0000
- title: 'Do ImageNet Classifiers Generalize to ImageNet?'
abstract: 'We build new test sets for the CIFAR-10 and ImageNet datasets. Both benchmarks have been the focus of intense research for almost a decade, raising the danger of overfitting to excessively re-used test sets. By closely following the original dataset creation processes, we test to what extent current classification models generalize to new data. We evaluate a broad range of models and find accuracy drops of 3% - 15% on CIFAR-10 and 11% - 14% on ImageNet. However, accuracy gains on the original test sets translate to larger gains on the new test sets. Our results suggest that the accuracy drops are not caused by adaptivity, but by the models’ inability to generalize to slightly "harder" images than those found in the original test sets.'
volume: 97
URL: https://proceedings.mlr.press/v97/recht19a.html
PDF: http://proceedings.mlr.press/v97/recht19a/recht19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-recht19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Benjamin
family: Recht
- given: Rebecca
family: Roelofs
- given: Ludwig
family: Schmidt
- given: Vaishaal
family: Shankar
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5389-5400
id: recht19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5389
lastpage: 5400
published: 2019-05-24 00:00:00 +0000
- title: 'Fast Rates for a kNN Classifier Robust to Unknown Asymmetric Label Noise'
abstract: 'We consider classification in the presence of class-dependent asymmetric label noise with unknown noise probabilities. In this setting, identifiability conditions are known, but additional assumptions were shown to be required for finite sample rates, and so far only the parametric rate has been obtained. Assuming these identifiability conditions, together with a measure-smoothness condition on the regression function and Tsybakov’s margin condition, we show that the Robust kNN classifier of Gao et al. attains, the mini-max optimal rates of the noise-free setting, up to a log factor, even when trained on data with unknown asymmetric label noise. Hence, our results provide a solid theoretical backing for this empirically successful algorithm. By contrast the standard kNN is not even consistent in the setting of asymmetric label noise. A key idea in our analysis is a simple kNN based method for estimating the maximum of a function that requires far less assumptions than existing mode estimators do, and which may be of independent interest for noise proportion estimation and randomised optimisation problems.'
volume: 97
URL: https://proceedings.mlr.press/v97/reeve19a.html
PDF: http://proceedings.mlr.press/v97/reeve19a/reeve19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-reeve19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Henry
family: Reeve
- given: Ata
family: Kaban
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5401-5409
id: reeve19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5401
lastpage: 5409
published: 2019-05-24 00:00:00 +0000
- title: 'Almost Unsupervised Text to Speech and Automatic Speech Recognition'
abstract: 'Text to speech (TTS) and automatic speech recognition (ASR) are two dual tasks in speech processing and both achieve impressive performance thanks to the recent advance in deep learning and large amount of aligned speech and text data. However, the lack of aligned data poses a major practical problem for TTS and ASR on low-resource languages. In this paper, by leveraging the dual nature of the two tasks, we propose an almost unsupervised learning method that only leverages few hundreds of paired data and extra unpaired data for TTS and ASR. Our method consists of the following components: (1) denoising auto-encoder, which reconstructs speech and text sequences respectively to develop the capability of language modeling both in speech and text domain; (2) dual transformation, where the TTS model transforms the text $y$ into speech $\hat{x}$, and the ASR model leverages the transformed pair $(\hat{x},y)$ for training, and vice versa, to boost the accuracy of the two tasks; (3) bidirectional sequence modeling, which address the error propagation problem especially in the long speech and text sequence when training with few paired data; (4) a unified model structure, which combines all the above components for TTS and ASR based on Transformer model. Our method achieves 99.84% in terms of word level intelligible rate and 2.68 MOS for TTS, and 11.7% PER for ASR on LJSpeech dataset, by leveraging only 200 paired speech and text data (about 20 minutes audio), together with extra unpaired speech and text data.'
volume: 97
URL: https://proceedings.mlr.press/v97/ren19a.html
PDF: http://proceedings.mlr.press/v97/ren19a/ren19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ren19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Yi
family: Ren
- given: Xu
family: Tan
- given: Tao
family: Qin
- given: Sheng
family: Zhao
- given: Zhou
family: Zhao
- given: Tie-Yan
family: Liu
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5410-5419
id: ren19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5410
lastpage: 5419
published: 2019-05-24 00:00:00 +0000
- title: 'Adaptive Antithetic Sampling for Variance Reduction'
abstract: 'Variance reduction is crucial in stochastic estimation and optimization problems. Antithetic sampling reduces the variance of a Monte Carlo estimator by drawing correlated, rather than independent, samples. However, designing an effective correlation structure is challenging and application specific, thus limiting the practical applicability of these methods. In this paper, we propose a general-purpose adaptive antithetic sampling framework. We provide gradient-based and gradient-free methods to train the samplers such that they reduce variance while ensuring that the underlying Monte Carlo estimator is provably unbiased. We demonstrate the effectiveness of our approach on Bayesian inference and generative model training, where it reduces variance and improves task performance with little computational overhead.'
volume: 97
URL: https://proceedings.mlr.press/v97/ren19b.html
PDF: http://proceedings.mlr.press/v97/ren19b/ren19b.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-ren19b.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Hongyu
family: Ren
- given: Shengjia
family: Zhao
- given: Stefano
family: Ermon
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5420-5428
id: ren19b
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5420
lastpage: 5428
published: 2019-05-24 00:00:00 +0000
- title: 'Adversarial Online Learning with noise'
abstract: 'We present and study models of adversarial online learning where the feedback observed by the learner is noisy, and the feedback is either full information feedback or bandit feedback. Specifically, we consider binary losses xored with the noise, which is a Bernoulli random variable. We consider both a constant noise rate and a variable noise rate. Our main results are tight regret bounds for learning with noise in the adversarial online learning model.'
volume: 97
URL: https://proceedings.mlr.press/v97/resler19a.html
PDF: http://proceedings.mlr.press/v97/resler19a/resler19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-resler19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alon
family: Resler
- given: Yishay
family: Mansour
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5429-5437
id: resler19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5429
lastpage: 5437
published: 2019-05-24 00:00:00 +0000
- title: 'A Polynomial Time MCMC Method for Sampling from Continuous Determinantal Point Processes'
abstract: 'We study the Gibbs sampling algorithm for discrete and continuous $k$-determinantal point processes. We show that in both cases, the spectral gap of the chain is bounded by a polynomial of $k$ and it is independent of the size of the domain. As an immediate corollary, we obtain sublinear time algorithms for sampling from discrete $k$-DPPs given access to polynomially many processors. In the continuous setting, our result leads to the first class of rigorously analyzed efficient algorithms to generate random samples of continuous $k$-DPPs. We achieve this by showing that the Gibbs sampler for a large family of continuous $k$-DPPs can be simulated efficiently when the spectrum is not concentrated on the top $k$ eigenvalues.'
volume: 97
URL: https://proceedings.mlr.press/v97/rezaei19a.html
PDF: http://proceedings.mlr.press/v97/rezaei19a/rezaei19a.pdf
edit: https://github.com/mlresearch//v97/edit/gh-pages/_posts/2019-05-24-rezaei19a.md
series: 'Proceedings of Machine Learning Research'
container-title: 'Proceedings of the 36th International Conference on Machine Learning'
publisher: 'PMLR'
author:
- given: Alireza
family: Rezaei
- given: Shayan Oveis
family: Gharan
editor:
- given: Kamalika
family: Chaudhuri
- given: Ruslan
family: Salakhutdinov
page: 5438-5447
id: rezaei19a
issued:
date-parts:
- 2019
- 5
- 24
firstpage: 5438
lastpage: 5447
published: 2019-05-24 00:00:00 +0000
- title: 'A Persistent Weisfeiler-Lehman Procedure for Graph Classification'
abstract: 'The Weisfeiler–Lehman graph kernel exhibits competitive performance in many graph classification tasks. However, its subtree features are not able to capture connected components and cycles, topological features known for characterising graphs. To extract such features, we leverage propagated node label information and transform unweighted graphs into metric