Hi, I'm attempting a refactoring of the SharpNEAT code base, converting the code to use generics (The dotNet world's name for templates) and generally just...
... That is interesting, I had not thought of that. It would be interesting to look at patterns generated by such a CPPN. ... What did you mean by playing...
Dear all, I am just trying to understand... The proposal is to apply GNG to define the hidden node structure in a hyperneat substrate? One CPPN (4D) is still...
Yes, exactly. In fact this can be applied to not only HyperNEAT neural substrates, but generally any 2D/3D mesh structure. The CPPN can still take only 4...
Here are some thoughts on speciation that probably don't warrant a deep discussion but I post them for the record. Currently sharpneat takes a casual approach...
Here's a speciation/clustering method that may by useful for speciating initial populations or respeciating when the number of species deviates from the target...
Hey Colin, You have some interesting ideas for refining speciation. I wonder what sort of performance impact these changes might have. I wanted to clarify your...
Colin, I read through both your posts on speciation. There is definitely room for innovation on the way that most NEAT implementations (including my own)...
Oops, looks like Joel and I wrote the same clarification on novelty search at the same time :) Oh well, Joel's is more informative than what I wrote. ken...
Hi Ken, ... the ... This issue is something I was thinking of very often. Because one species can be defined as a group of genomes that have many matching ...
Peter, yes, that is another way that it could be done that I forgot to mention: You could characterize species by some kind of consolidated statistic like...
I agree, Ken. One example cheap solution for good stability is ensuring that the compatibility threshold is adequate to the limits of the system. Maybe...
Hi all, Thanks for the feedback so far. There are a few issues raised and I'll try to cover all of them. Apologies for the length of the post. Firstly I did of...
Hi Colin, ... calculating ... is ... genomes ... local ... our ... I agree. I think you mean that "a hole" is possible, and averaging the weights like that...
Hi, ... Yes and no. Consider the outer boundary of a species defined by the genomes at the extremeties. Now consider the enclosed space as a mass with uniform...
I uploaded a version of Mat's NEAT program that will compile under Visual Studio 2005. All I did was fix in-loop variable creations (a no- no from 2005 onward)...
Colin, I finally got around to reading through your long post reasoning about speciation techniques. I appreciate the overview of pros and cons of several...
Lancer, thanks for the contribution. Several years ago, Prof. Stephanie Forrest at the University of New Mexico told me about a class assignment she had given...
Hello all- Ever since I first understood HyperNEAT I have wanted to see how it would perform at controlling legged robots. A paper I wrote on that subject was...
Cícero Augusto Bacharel em Ciência da Computação - UFBA Mestrando em Ciência da Computação - UFSC ... "You've achieved success in your field when you...
Jeff, Cool! Thanks for sharing the paper and video's. I've given the paper a quick read, but I really need to spend more time on it. I hope to be able to...
I've been thinking of adding CUDA support to my HyperNEAT implementation. Can you elaborate on the process of porting the code to the GPU? Did you find any...
Hi Jeff. Very interesting paper! :) It reminds me about the experiments I made with HyperNEAT for controlling biped robots. Did you try using leaky-integrator...
... Thanks. I'm glad you enjoyed it. ... I have thought a lot about your biped experiment. I'd be really interested in hearing your opinion, and that of anyone...
Hi Jason, The process of porting the code is an interesting (to me) discussion. The central issue (some might call pitfall) is partitioning for parallel ...
... Hi Ken. Thanks for checking it out. What did you think of the videos? I am interested to see if others agree with my feelings that the HyperNEAT robotic...
Hi Jeff, Thanks for posting the paper, nice work. It's another clear win for HyperNEAT and in an area with commercialisation opportunities I would suggest. ...
Hi Jeff. Here is my opinion about the 4 hypothesis you suggested. ... Definitely true! :) While robots with 3+ legs are generally stable, biped robots have...
Just a random thought I had about evolving biped gaits: Have you considered first evolving a quadruped gait, then switching the fitness function to one that's...
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.
