Dear Shimon, In a previous message Kenneth noted that in robots and robot simulators the inputs and outputs are not necessarily in a format that is natural or...
Take a look at Torque. http://www.garagegames.com/ the engine was designed for a fps. But I have worked with it some and there is a huge community that uses it...
Chris, thanks for the suggestion. Actually we are using Torque right now, but the problem is Torque is not really geared towards RTS-style games. Torque was...
German, Ken's suggestions sound good to me. In our project, Nate and I did not consider these kind of inputs representations so I don't have any data that can...
... This sounds very similar to the results we got before we started using task decompositions. I think the problem is that soccer is a task with *very* ...
... [ snipped long description of proposed algorithm ] ... In a second... ... That was my original idea... ... Ah-ha! That's what I wanted to know. So then...
Hi John, Have you applied this approach sucessfully to the image enlargement domain? I think innovation may come from the low ranking specimens (for example...
Ok I think I'm understanding now, and I think it makes sense, as long as NEAT isn't involved in the evaluation step. That is, if I am understanding correctly,...
... Yep. I might ask NEAT to get involved in figuring out if it is 'enough' or not, or I may just duplicate that code within the experiment. Whichever is ...
From the beginning, I was concerned about one thing: The output is a fixed multiplicative factor. I didn't like that. I suspect that deep in the bowels of...
Thanks to everyone who suggested an RTS game engine. Unfortunately we really need something closer to commercial quality, so we may end up having to either...
I put up code on the NEAT Users page under the question: http://www.cs.utexas.edu/users/kstanley/neat.html#disconnected There is a link to code for JNEAT, and...
Hi John, As far as evolving a parameterized/continous enlargment NN goes I think you're on the right lines but I'm a bit concerned that you are trying to 'jump...
John, Colin makes a good point that this method greatly complicates the function to be computed. But on the other hand, as you pointed out, it does make the...
As some of you will be aware I have been working on a C# implementation of NEAT (which I have deciced to name SharpNEAT). The final pieces of the core code...
Colin, that sounds like a promising start! How many generations did you run, what was your pop size, and did you get a chance to look at the winning structure...
... to ... run...huge ... popsize was 150. The first few easy tasks were solved within the first dozen generations - which is behaviour I had witnesses before....
... It is a good point. But I'm not sure I agree with 'greatly.' Yes, it is more complicated. However, having a fixed enlargement amount and fixed output...
... did ... at ... useful ... before. ... generations. ... This sound like a bit too many species for a population of 150. Ideally, you'd want around 10...
... fitness ... Perl is ... John, does this mean you will be using JNEAT? Or are you going to write your own, or perhaps use another Java version of NEAT? ken...
... OK I need to look into this. I'm not sure what the spread across the species was - but I agree it seems a high number. ... Sure. This is basically the...
... I'm thinking hybrid. I want to do a more passive system, like SETI@home, where clients ping the server to get work to do. That way, I can put computers...
482
Michael Gillam
gillam@...
Mar 9, 2004 1:22 am
Hey Ken, If you decide to use a FPS engine instead of an RTS engine you may wish to contact ID Software. From our experience, they can be pretty open about ...
... [Sound of me climbing onto soapbox] The .Net framework and associated languages are also a possibility for multi-platform stuff thanks to the Mono project,...
Making NEAT distributed definitely sounds like a good idea and should be relatively straightforward. I'm wondering, has anyone else here tried or done any...
I've come against a small problem with the fitness sharing technique so I would like to ask how others are handling this... I started by calculating the size...
Mike, thanks for the suggestion on ID. It's good to know and I'll keep it in mind if we decide FPS would be suitable in the long run. ... may wish ... open...
... and ... hack ... occurs. ... I've been using #2. Actually, I think my version of NEAT will change any fitness of 0 to 0.000001 or something like that. I...
Hi I don't know if this forum is also for technical quistions about the C++ implementation but here is my problem: I am trying to compile the version without...
