Wesley R. Elsberry
Posts: 4991 Joined: May 2002
|
Quote (GaryGaulin @ Mar. 12 2015,21:35) | Quote (Wesley R. Elsberry @ Mar. 12 2015,21:02) | Quote (GaryGaulin @ Mar. 12 2015,20:44) | Quote (Wesley R. Elsberry @ Mar. 12 2015,08:28) | Quote (GaryGaulin @ Mar. 12 2015,06:40) | Quote (Wesley R. Elsberry @ Mar. 11 2015,17:17) | Gary:
Quote | I know you cannot prove the claims that you and others make in regards to Avida models.
|
Interesting. I have a paper on my use of Avida, and I'm sure that we backed up our claims, so that's another of Gary's claims that turns out to be FALSE. |
Show everyone how you objectively detect, qualify and quantify "intelligence". Where is the basic ability to take a "guess" located in their brain and how does the subsystem work? |
That's not how it works. I'm on the hook to support the claims that I made, not the ones Gary wants to "Lets Pretend!" I made.
Show me a claim about Avida that I made that I didn't back up, Gary.
Or Gary could show us line numbers for the implementations of the various things that are claimed to be in his code, or that ought to be in his code if Gary had a clue about the stuff he was trying to "model". Where is the stuff handling multiple spatial scales? Or noise in the system? Or the phase considerations? Or the connectivity patterns aligned with spatial phase, the issue noted by two of the three Nobel laureates Gary referenced earlier as crucial to the validity of an attractor network model? None of those was apparent on inspection of GridCellNetwork.frm. |
You have been claiming that Avida is for "evolving intelligence" now back up your own damn claims!!!!!!!
From: http://www.kurzweilai.net/forums.....-692448 Quote | RBynum, there are a number of interesting systems that are variants of genetic programming. You might be especially interested in the field of artificial life.
I did some work with the Avida system. Avida's digital organisms, or Avidians, are programs constructed with a Turing-complete instruction set. I took part in Rob Pennock's "evolving intelligence" project. My particular interest was to extend Avida to do some investigation of movement. I added another three instructions to the Avida language to permit Avidians to move about a grid. One allowed the Avidian to sense the difference in resource concentration between its cell and the cell it was facing. Another allowed it to turn to a new facing randomly ("tumble"). And the last actually allowed it to move to the cell it faced ("move"). The limitation on the sensing to strictly local information meant that there was no way for the Avidian to have prior knowledge of a distant resource peak.
The seed organism in an experimental run was an Avidian that could perform self-replication, but no other task. In my experiments, I used a resource in high concentration in a relatively small part of a bounded grid, and a gradient of the resource in weaker concentration across the rest of the grid. In a great many experimental runs, what I observed was that Avidians would evolve a number of classes of movement strategies. One of those classes was that of gradient ascent programs. I was also able to use the evolved Avidian programs as robot control programs, where I used a single light bulb as the "resource" and light sensors on the robot for the sensory system. (I used both iRobot Create and Lego MindStorms systems.)
Paper: Cockroaches, Drunkards, and Climbers
A colleague of mine, Laura Grabowski, did experiments with Avidians evolving to follow complex paths. I think that you would find her work interesting, too. |
I modeled the common "ANN" that's also found in Avida and found that such networks are NOT intelligent. In fact I was astonished by how unintelligent it is. Now explain how YOU tested the Avida model for the presence of brain produced intelligence. Or did you just blindly believe what you were told and now get snotty whenever someone questions your beliefs? |
And I have already documented the support for my claims with respect to Avida. I've even linked to that PDF before, so Gary doesn't even have the excuse of ignorance for his FALSE claim.
Gary claims that he 'modeled the common "ANN" that's also found in Avida'. Since Avida contains no artificial neural network, here we have yet another of Gary's claims that is simply FALSE. |
From the Avida website:
Where is this "evolving intelligence" to be found in any Avida anywhere? I see you and others making the claim. Now back it up by showing me where it is! |
I've already provided the link supporting my claims, making Gary's assertion that this is yet to be done yet another claim that is FALSE.
Gary makes much about copying a reference to a Pennock paper from the "Avida" website. A minor FALSE claim from Gary, who didn't get that link from the Avida website; he got it from the Avida-ED website. N. Wells is right that the paper in question has nothing to do with what is in Avida, and everything to do with the capabilities of evolutionary computation taken broadly. Gary even quoted the lead-in line that states that the reference is about evolutionary computation. Careful readers (i.e., someone not named "Gary S. Gaulin") have no trouble taking the cue that the topic of the paper is not Avida-specific, or actually reading the paper itself and finding out what the content is before trying to characterize it. So Gary is still on the hook for the FALSE claim that Avida has an "ANN" in it. Now, given that Gary obviously hadn't even read the paper he claimed as support for there being an "ANN" in Avida, there is the psychological problem of what, exactly, was the nature of the hallucination Gary had about that. Gary could, I suppose, pony up his code where he supposedly made this stunning discovery so that we all can figure out exactly what delusional "Let's Pretend!" process he was following, but I am not expecting Gary to do anything to rebut the impression that he has made yet another claim that can be nothing except FALSE.
-------------- "You can't teach an old dogma new tricks." - Dorothy Parker
|