JT, thanks for the upload. If you don't feel it's polished enough it may not be the best idea to release it widely (i.e. from our HyperNEAT page) since it...
Hi Ken I had done an experiment in conjuction with the Torcs competition that Julian Togelius et al are running. The results were not that encouraging when...
Hi Matt, Thanks for pointing out this inconsistency. You're right that both papers managed to somehow list different but still incorrect fitness functions. I...
Matt, please let me also apologize for the typos in the fitness function in the food gathering experiment publications. We try to be vigilant about the...
Matt, you mentioned attempting to apply HyperNEAT to the Torcs competition, but not doing better than NEAT. Perhaps it is indeed the geometry that could be...
Hello Ken- I agree with you that, from what I have read and personally experienced, evolution likes to evolve static, rather than adaptive, heuristics. It...
Hi everyone, I've started a small GPL C# demonstration program on SourceForge to show how a Factory Pattern might be used to generate and compile HyperNEAT on...
Ken, that looks like an interesting project, thanks for sharing it. I was curious, is the C# implementation of HyperNEAT within the CUDA project your own, or...
Jeff, my hypothesis is that static heuristics are easier to find than adaptive ones because it is a lot easier to encode knowledge than to encode how to gain...
Ken, Currently, the project is based on David D'Ambrosio's HyperSharpNeat v2.0. That library will undergo some major changes to be compatible with CUDA, but I...
What are you planning to parallelize in CUDA? The bottleneck in most of the experiments in neuroevolution is the fitness function, which is very specific for...
Cesar, I've found many bottlenecks in the partitioning schema for various HyperNEAT experiments. They migrate through the compute fabric, like constraints in ...
Hi NEAT-ers, just wanted to remind you of the next CIG conference. (I know several of you are already planning to submit and attend.) Cheers, Julian IEEE...
By the way, about the CIG Symposium, I will be giving one of the tutorials there: TUTORIALS (from Julian's announcement) * Playing Machines: Machine Learning...
... I look forward to that tutorial! I read the slides of the previous tutorial, and it's very important stuff that's rarely treated in its own right in...
A review article could be a good idea but the problem is that the subject matter is not very scientific. It's more practical/technical. That's why I have...
Hi, I am porting NEAT to Nvidia GT120 based game cards (GTX 260+ series) with CUDA. The problem is, the GPU executes double precision operations 8 times slower...
Nikolai, What I have done in my SourceForge implementation is simply convert Colin Green / David D'Ambrosio's HyperNEAT standard library to floats. I have ...
For the record the default neural net implementation in SharpNeat uses single precision floats although there are implementations that use double, and also the...
Colin, Thank you for this background information. In many gradient descent (or hill climbing), there are two very real problems. The first is becoming stuck...
Ken (L), ... I've not thougth of SA as a means to overcome low perturbation potential (as you put it), but yes I can see how it would help. I do get the sense...
Hey Nikolai, My C++ implementation of HyperNEAT has a typedef that makes it easy to change between double and single precision. I've tested both with many of...
Colin and the group, I feel like I need to re-think my objectives for HyperNEAT4CUDA. Originally, my thought was a simple demonstration means of using a...
Hi everybody, Thought this might be of interest to some of you on the list working in Evolutionary Robotics: We are organizing a half-day IROS workshop...
Hi, ... I don't really follow, is this some scheme to chose node coordinates on the hyperneat substrate rather than using fixed points. ... To me this...
Colin, and other interested members, Perhaps you might join the Human-Agent-Robot-Teamwork (HART) workshop next year. Or, join the Agent Based Modeling and...
Hi, i got few errors with rtneat on gcc v4. under gcc v3 it compiles fine. Some tuning must be done in order to compile correctly for gcc 4, does someone has...
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.
