The beneficial objectives of Supervisor NEAT are twofold: a) Enable implementation of pyramid number of populations vs. starting genome complexity; and b)...
Hi there, I am using a CTRNN version of rtNEAT, which is an upgraded version of Tyler Streeter's leakyNEAT code, combined with the rtNEAT release. The problem...
Hi, Three newbie questions: 1) As I understand it, (fully connected) NEAT starts out with just one bias node - is that right? 2) Would there be any advantage...
... NEAT only ever uses one bias node. There's no point in having more than one. ... I don't know, but I doubt, you'd find any advantage in evolving node ...
Steve, thanks for providing those answer, I agree with what you said. About the generation zero vs. one issue, of course ultimately it doesn't matter what you...
Petar, CTRNNs can be really hard to callibrate right to get them working at first. They are very sensitive to things like the range of time constants, range...
... Oh yes, sorry about that.. I assumed that was obvious. I mistook the question to be simply a matter of what you should name the first generation (0 or 1)....
... Steve and Ken Actually it was my simple question about 0 or 1. Of course I could have just looked at the existing code in the different implementations,...
Hi all, I'm a grad student at Virginia Tech and I would like to enter a bot into the AAAI poker competition in July. I've been working on poker AI in my spare...
I have developed much of my life to poker AI over the last year and half and, much as in other areas of AI and EC (evo comp), have been severely humbled. I...
... not fold ... I know the feeling. I've devoted a large portion of my life to poker as well, and have had less-than-stellar results in terms of a full-out ...
Just as a data point, I wrote a NEAT engine in C++ a few years back for the purposes of eventually teaching it poker. I too was humbled by the poker aspect...
... still ... I think we can do this if we are able to evolve separate networks in each individual. As I mentioned before, by picking an entire play at the...
... I think this is a neat idea (no pun) but I'm not sure that it ultimately helps. In my mind it seems that in this scenario, either we evaluate each network...
... That's true in a way. I don't really know which is slower, evolving multiple sets of similar nets or evolving one monolithic net and using separate...
... But wouldn't that break up the network's knowledge of what happened in previous rounds? Or would the turn also have an input that says what happened...
One thing I would suggest: start simple. I think the original model had a whole bunch of ideas about player modeling (which is the reason to include how...
And, ps., 3 is really hard! Cheers, Jeff Clune Digital Evolution Lab, Michigan State University Visiting Scholar, Instituto de Biología Molecular y Celular de...
... agreed. ... Not to sound like a stuck record but I still wonder if something faster wouldn't be possible and maybe necessary. One thing I started to...
... The latter, as I described in either my original or second post on this thread. The point of having separate networks is to be able to have a fixed number...
... I'm not sure I completely agree with the assumption that we should be representing the table "literally". I feel like there's simply too much information...
About KB-NEAT, it might be useful for poker. Ryan Cornelius is the one who coded KB-NEAT, and he even had it working on blackjack, I believe. But I don't...
... Sorry Jeff, my 4am emails don't really get the point across very well. When I say having a simulation engine, I am talking about one that provides...
... I see...well, it's hard to say insofar as I don't know what that information is right now. That is to say, I'm still up in the air in my own mind as to...
Hi Guys, I am currently working on evolving a poker bot using a parallel genetic algorithm evolving neural networks under java. I'm not using neat yet as I...
I'm not quite sure though if just the probability of winning vs a random opponent is enough information about the hand. I assumed it's all one needs but maybe...
... no, not even close. I was just giving it as an example of information that may be useful (or more useful) beyond just the raw data (e.g. the cards in your...
We are running a simulated car racing competition as part of the IEEE Symposium on Computational Intelligence and Games. One motivation for this competition is...
Having problems with message search?
Fill out this form to ensure your group is one of the first to be migrated to the new message search system.
