Antievolution.org :: Antievolution.org Discussion Board The Critic's Resource on Antievolution

Antievolution.org Discussion Board > From the Panda's Thumb > After the Bar Closes... > IS empiricism a natural part of pattern recognitio

Pages: (3) < 1 2 [3] >

sledgehammer Posts: 533 Joined: Sep. 2008 (Permalink) Posted: Mar. 03 2010,12:03    Quote (BWE @ Mar. 03 2010,08:42) In the last 5 years, the world changed. Learning algorithms are commonplace, there are dozens of computer models of neural networks which function exactly as they would be expected to, and embodied intelligence has crossed the hurdle of creativity. If you want a good place to start, http://www.umass.edu/neuro/faculty/files/siegelmann.html EDIT: No link on that page to her publications. If you actually care, here is that link: http://binds.cs.umass.edu/publications.html this lady has some interesting math she has been working on for quite some time. The list however, is nearly endless. Impressive. The work she's doing is real bioinformatics, with practical applications to real-world problems like addiction and circadian responses.  Her "Active Information" is on a completely different level from the crap produced by the Evolutionary Informatics "lab" of Dembski, Marks, Sewell and Dodgen. Regards to randomness, from her 1998 paper "Neural Dynamics With Stochasticity"   Quote Randomness is a basic characteristic of large distributed systems. It may result from the activity of the individual agents, from unpredictable changes in the communication pattern among the agents, or even just from the different update paces ... Our particular stochastic model can be seen as an incorporation of the von Neumann model of unreliable interconnections of components to the area of neural networks: the basic component has a fixed probability for malfunction. The last is a particularly apt description of my brain, except for the "fixed probability" part   -------------- The majority of the stupid is invincible and guaranteed for all time. The terror of their tyranny is alleviated by their lack of consistency. -A. Einstein  (H/T, JAD) If evolution is true, you could not know that it's true because your brain is nothing but chemicals. ?Think about that. -K. Hovind

BWE Posts: 1902 Joined: Jan. 2006 (Permalink) Posted: Mar. 03 2010,12:07    Quote (midwifetoad @ Mar. 03 2010,09:34) Quote there are dozens of computer models of neural networks which function exactly as they would be expected to My point when I said quantum consciousness solves a problem not in evidence. In order to have a need for such a hypothesis, you need a recalcitrant problem. Simply not knowing how self-awareness arises is not such a problem. The cost of hardware is a problem, which is why I'm interested in Stanford. A) You are right. There is no problem for self-awareness that quantum issues raise. There may be quantum issues but they appear to resolve at our particular recursive level before we need to know about them much. Which is why Penrose doesn't understand the opposition to his hypothesis. He is simply working out a detail of extraordinarily esoteric math unless someone notices a problem that needs fixing. B) Stanford? -------------- Who said that ev'ry wish would be heard and answered When wished on the morning star Somebody thought of that, and someone believed it Look what it's done so far The Daily Wingnut

BWE Posts: 1902 Joined: Jan. 2006 (Permalink) Posted: Mar. 03 2010,12:12    Quote (sledgehammer @ Mar. 03 2010,10:03) Quote (BWE @ Mar. 03 2010,08:42) In the last 5 years, the world changed. Learning algorithms are commonplace, there are dozens of computer models of neural networks which function exactly as they would be expected to, and embodied intelligence has crossed the hurdle of creativity. If you want a good place to start, http://www.umass.edu/neuro/faculty/files/siegelmann.html EDIT: No link on that page to her publications. If you actually care, here is that link: http://binds.cs.umass.edu/publications.html this lady has some interesting math she has been working on for quite some time. The list however, is nearly endless. Impressive. The work she's doing is real bioinformatics, with practical applications to real-world problems like addiction and circadian responses.  Her "Active Information" is on a completely different level from the crap produced by the Evolutionary Informatics "lab" of Dembski, Marks, Sewell and Dodgen. Regards to randomness, from her 1998 paper "Neural Dynamics With Stochasticity"     Quote Randomness is a basic characteristic of large distributed systems. It may result from the activity of the individual agents, from unpredictable changes in the communication pattern among the agents, or even just from the different update paces ... Our particular stochastic model can be seen as an incorporation of the von Neumann model of unreliable interconnections of components to the area of neural networks: the basic component has a fixed probability for malfunction. The last is a particularly apt description of my brain, except for the "fixed probability" part  :p This one here: http://binds.cs.umass.edu/papers/2008_Olsen_AI.pdf Is one which I've written a little about in other places.  But yes. This is the actual state of the discipline. The issue of quantum consciousness is outdated. The real research has moved on and, like I said, AI is already a reality. Within a decade we will likely have self-awareness from a silicone base. The difference between now and 5 years ago is the difference between Einstein and Faraday. -------------- Who said that ev'ry wish would be heard and answered When wished on the morning star Somebody thought of that, and someone believed it Look what it's done so far The Daily Wingnut

midwifetoad Posts: 4003 Joined: Mar. 2008 (Permalink) Posted: Mar. 03 2010,13:56    Quote B) Stanford? http://www.stanford.edu/group/brainsinsilicon/goals.html As far as I can tell, the innovation here is one of cost and power consumption. Quote This neuromorphic approach, developed over the past two decades, yields hitherto unimagined levels of efficiency that make Blue-Gene performance affordable on a Dell-cluster budget. -------------- Any version of ID consistent with all the evidence is indistinguishable from evolution.

BWE Posts: 1902 Joined: Jan. 2006 (Permalink) Posted: Mar. 03 2010,14:18    Wow. That's pretty amazing. I just re-wrote a reply enough times that I think I need to wait a bit to think about it for a few minutes. -------------- Who said that ev'ry wish would be heard and answered When wished on the morning star Somebody thought of that, and someone believed it Look what it's done so far The Daily Wingnut

Thought Provoker Posts: 530 Joined: April 2007 (Permalink) Posted: Mar. 03 2010,16:41    I have been looking into Siegelmann's published papers. Thank you BWE, they are very interesting. I was having trouble finding a specific disagreement between Siegelmann and Penrose but I finally found this... Quote Our model may also be thought of as a possible answer to Penrose’s recent claim Penrose (1989) that the standard model of computing is not appropriate for modeling true biological intelligence. Penrose argues that physical processes, evolving at a quantum level, may result in computations which cannot be incorporated in Church’s Thesis. The analog neural network does allow for non-Turing power while keeping track of computational constraints, and thus embeds a possible answer to Penrose’s challenge within the framework of classical computer science. link It seems Siegelmann is relying on the analog part of "analog neural network" to get to her AI nirvana.  This doesn’t make a lot of sense to me because, as an Electrical Engineer, I work with analog-to-digital conversion with precisions approaching electron charge.  Penrose is a mathematician.  Reducing digital precision by orders of magnitude is still digital precision.  Penrose can be wrong, but I don’t see how analog verses digital changes anything in our quantum-based reality. Let me cut to what I see as fundamental differences.  I suspect most of you would agree it is possible life evolved to use quantum effects but you may argue it is just not required and, therefore, can be simulated "...within the framework of classical computer science."   This gets into the discussion about "sources of true randomness" postulated by sledgehammer.  If quantum effects is the fundamental source for all true randomness then it becomes a metaphysical free-for-all. Is it truly random? Is it orchestrated quantum effects in spacetime? Is it God working in "mysterious ways"?

BWE Posts: 1902 Joined: Jan. 2006 (Permalink) Posted: Mar. 03 2010,17:17    Quote (Thought Provoker @ Mar. 03 2010,14:41) I have been looking into Siegelmann's published papers. Thank you BWE, they are very interesting. I was having trouble finding a specific disagreement between Siegelmann and Penrose but I finally found this... Quote Our model may also be thought of as a possible answer to Penrose’s recent claim Penrose (1989) that the standard model of computing is not appropriate for modeling true biological intelligence. Penrose argues that physical processes, evolving at a quantum level, may result in computations which cannot be incorporated in Church’s Thesis. The analog neural network does allow for non-Turing power while keeping track of computational constraints, and thus embeds a possible answer to Penrose’s challenge within the framework of classical computer science. link It seems Siegelmann is relying on the analog part of "analog neural network" to get to her AI nirvana.  This doesn’t make a lot of sense to me because, as an Electrical Engineer, I work with analog-to-digital conversion with precisions approaching electron charge.  Penrose is a mathematician.  Reducing digital precision by orders of magnitude is still digital precision.  Penrose can be wrong, but I don’t see how analog verses digital changes anything in our quantum-based reality. Let me cut to what I see as fundamental differences.  I suspect most of you would agree it is possible life evolved to use quantum effects but you may argue it is just not required and, therefore, can be simulated "...within the framework of classical computer science."   This gets into the discussion about "sources of true randomness" postulated by sledgehammer.  If quantum effects is the fundamental source for all true randomness then it becomes a metaphysical free-for-all. Is it truly random? Is it orchestrated quantum effects in spacetime? Is it God working in "mysterious ways"? TP, The thing about Penrose's hypothesis that makes it a side issue is that AI research is not hampered by it. I noted Siegelmann because she has had some recent sort of spectacular successes but the list is long. The direction for now seems clear in AI research and snags thus far aren't appearing. There are two different questions involved in Penrose's issue, one biological and the other computational. Neuroscience and AI are beginning to come together more and more to bridge the two, but the questions of what human consciousness is and what self awareness in general is are easy to separate by discipline. Trying to arrive at a definition which excludes artificial consciousness from the domain of human consciousness is the next step typically. AI researchers can largely ignore that one because their goal is not to answer the question but to create consciousness in a machine. Also, mathematical continuums are fundementally not the same as physical continuums. A physical continuum, no matter how good your measurement system, always has the problem of measurement where at some point, A=B, B=C and A<C. Zeno's paradox is answered in a physical continuum because at some point achilles ends up at B. This is a major difference between digital and analog signal processing. Not that Siegelmann's work is the only method out there, there are several digital efforts that appear to be having success too. The important thing about them all is that Penrose's objection is not stopping any of them. It doesn't even need to invoke 'true' randomness, whatever that is. -------------- Who said that ev'ry wish would be heard and answered When wished on the morning star Somebody thought of that, and someone believed it Look what it's done so far The Daily Wingnut

Thought Provoker Posts: 530 Joined: April 2007 (Permalink) Posted: Mar. 03 2010,20:12    Quote (BWE @ Mar. 03 2010,17:17) The thing about Penrose's hypothesis that makes it a side issue is that AI research is not hampered by it. I suggest Penrose's hypothesis is not a "side issue" for understanding Quantum Mechanics.   Quote (BWE @ Mar. 03 2010,17:17) The direction for now seems clear in AI research and snags thus far aren't appearing. Personally, I hope AI research continues at a brisk pace because I suspect it will eventually bring to light what Penrose has been saying.   Quote (BWE @ Mar. 03 2010,17:17) Also, mathematical continuums are fundementally not the same as physical continuums. A physical continuum, no matter how good your measurement system, always has the problem of measurement where at some point, A=B, B=C and A<C. Zeno's paradox is answered in a physical continuum because at some point achilles ends up at B. This is a major difference between digital and analog signal processing. The difference between a mathematical, digital continuum and a physical, analog continuum is noise. Or putting it another way, uncertainty... ...as in quantum uncertainty. In electronics there are these things called "tunneling diodes".  Electrons manage to go from point A to point B without travelling in-between. Achilles can win a foot race with a tortoise because he teleports moment to moment.   Quote (BWE @ Mar. 03 2010,17:17) It doesn't even need to invoke 'true' randomness, whatever that is. A pseudo-random number generator is just a complex digital calculation.  The question is whether or not a purely digital machine is capable of AI consciousness.  From what I read of Siegelmann and other places, it is probable AI researchers are already giving up on what Penrose calls “Strong AI”.  The researchers are looking at quantum computations or analog signals with built in quantum noise.

BWE Posts: 1902 Joined: Jan. 2006 (Permalink) Posted: Mar. 03 2010,20:40    Quote (Thought Provoker @ Mar. 03 2010,18:12)     Quote (BWE @ Mar. 03 2010,17:17) The thing about Penrose's hypothesis that makes it a side issue is that AI research is not hampered by it. I suggest Penrose's hypothesis is not a "side issue" for understanding Quantum Mechanics. That is probably true. No doubt he has and will continue to add a lot to our understanding of various kinds of physics.     Quote     Quote (BWE @ Mar. 03 2010,17:17) The direction for now seems clear in AI research and snags thus far aren't appearing. Personally, I hope AI research continues at a brisk pace because I suspect it will eventually bring to light what Penrose has been saying. Well, suspect all you want. So far, it looks like a pretty well done deal and no problems reported thus far involving godel's theorems which, after all, are what Penrose bases his entire hypothesis upon.     Quote       Quote (BWE @ Mar. 03 2010,17:17) Also, mathematical continuums are fundementally not the same as physical continuums. A physical continuum, no matter how good your measurement system, always has the problem of measurement where at some point, A=B, B=C and A<C. Zeno's paradox is answered in a physical continuum because at some point achilles ends up at B. This is a major difference between digital and analog signal processing. The difference between a mathematical, digital continuum and a physical, analog continuum is noise. That is a ridiculous statement. No offense intended but if you think the only difference between the map and the terrain is noise you are going to need to take several steps backwards to get back on track.     Quote Or putting it another way, uncertainty... ...as in quantum uncertainty. or putting it another way, mathematical continuums are in no way whatsofuckingever the same as physical continuums. One is the map the other is the landscape.     Quote In electronics there are these things called "tunneling diodes".  Electrons manage to go from point A to point B without travelling in-between. Achilles can win a foot race with a tortoise because he teleports moment to moment. perhaps you can ruminate on this and discover why the Planck length is so fundamental to mathematical descriptions of a thing which isn't math.     Quote     Quote (BWE @ Mar. 03 2010,17:17) It doesn't even need to invoke 'true' randomness, whatever that is. A pseudo-random number generator is just a complex digital calculation.  The question is whether or not a purely digital machine is capable of AI consciousness.  From what I read of Siegelmann and other places, it is probable AI researchers are already giving up on what Penrose calls “Strong AI”.  The researchers are looking at quantum computations or analog signals with built in quantum noise. Correction, what John Searle calls "strong AI" and Penrose adopts. I believe I have discovered where our ideas diverge and which bits we differ over. Google "Chinese Room" and see if you think it's a reasonable analogy. -------------- Who said that ev'ry wish would be heard and answered When wished on the morning star Somebody thought of that, and someone believed it Look what it's done so far The Daily Wingnut

midwifetoad Posts: 4003 Joined: Mar. 2008 (Permalink) Posted: Mar. 04 2010,09:38    Quote Google "Chinese Room" and see if you think it's a reasonable analogy. Sorta like the difference between real living things and the hypothetical artificial living things that are the goal of biogenesis researchers. No? -------------- Any version of ID consistent with all the evidence is indistinguishable from evolution.

BWE Posts: 1902 Joined: Jan. 2006 (Permalink) Posted: Mar. 04 2010,10:12    Quote (midwifetoad @ Mar. 04 2010,07:38) Quote Google "Chinese Room" and see if you think it's a reasonable analogy. Sorta like the difference between real living things and the hypothetical artificial living things that are the goal of biogenesis researchers. No? touche. :) -------------- Who said that ev'ry wish would be heard and answered When wished on the morning star Somebody thought of that, and someone believed it Look what it's done so far The Daily Wingnut

midwifetoad Posts: 4003 Joined: Mar. 2008 (Permalink) Posted: Mar. 04 2010,11:18    I'm a bout done with this topic, but the problem I see with the Chinese Room argument is the assumption that emulations or silicon equivalents will be done via conventional programs running on conventional CPUs. I think Blue Gene makes a pretty good case that this will not happen in the real world. The cost is too high and the performance is too slow. It does, however, make the case that the behavior of neurons and neural networks can be understood, just as the chemistry of cells can be understood. Making high level artificial intelligence will require hardware breakthroughs in cost and efficiency (apparently this has begun). What you wind up with is not a CPU running a program, but a network of neurons that is functionally equivalent to biological neurons. You do not get a program that reads and writes Chinese via lookup tables. You get an evolvable network that can learn. I'm betting it will take far more than ten years to get something equivalent to an artificial human. After all, it takes more than ten years to raise a human, even having the hardware. But I'm betting there will be commercial applications of AI within ten years. Perhaps even decent non-colloquial language translation. Perhaps really good OCR and voice recognition. Perhaps devices that can listen to music, write  out a score, and perhaps generate a performance from a low-fi recording. -------------- Any version of ID consistent with all the evidence is indistinguishable from evolution.

Thought Provoker Posts: 530 Joined: April 2007 (Permalink) Posted: Mar. 04 2010,11:50    Hi BWE, I'm enjoying our conversation.  I hope you are too.  I'm trying not to be too frustrating but I want to explore our differences.  That usually involves a little provoking.  No offense intended or taken on my part.       Quote (BWE @ Mar. 03 2010,20:40)       Quote             Quote (BWE @ Mar. 03 2010,17:17) Also, mathematical continuums are fundementally not the same as physical continuums. A physical continuum, no matter how good your measurement system, always has the problem of measurement where at some point, A=B, B=C and A<C. Zeno's paradox is answered in a physical continuum because at some point achilles ends up at B. This is a major difference between digital and analog signal processing. The difference between a mathematical, digital continuum and a physical, analog continuum is noise. That is a ridiculous statement. No offense intended but if you think the only difference between the map and the terrain is noise you are going to need to take several steps backwards to get back on track. ... or putting it another way, mathematical continuums are in no way whatsofuckingever the same as physical continuums. One is the map the other is the landscape. It may be appropriate to bring the following into the conversation... For those unfamiliar with this famous painting... When is it the model of reality and when is it reality itself?  The Matrix movies explored this meme.  If Strong AI wins the day, then a conscious algorithm running inside a digital computer is just as real as a consciousness running inside an organic computer. If this digital computer also runs mathematical algorithms simulating reality, then for all intents and purposes it is a form of reality itself. Quantum Mechanical experiments pretty much show that our universe is a plank scale digital computer.  There is no such thing as particles, everything is a wavefunction (i.e. mathematical algorithm). So what differentiates this mathematical, digital continuum from the physical, analog continuum? You guessed it, noise in the form of quantum uncertainty.

Pages: (3) < 1 2 [3] >

Track this topic	Email this topic	Print this topic

© 2002-2009 Antievolution.org Powered by Ikonboard 3.0.2a © 2001 Ikonboard