Hi all, I'm proud to announce my first GECCO paper! Title: Accelerating Evolution via Egalitarian Social Learning Authors: Wesley Tansey, Eliana Feasley, and...
Hi Wesley, congrats on the GECCO paper and also thanks for sharing the software. I do have a couple questions/comments: The subculture idea is interesting but...
Hi Ken, Thanks. We did actually try leveraging the species created by NEAT, but unfortunately that wasn't useful for two reasons. First, the distribution of...
Just successfully defended my M.S. thesis work on Friday, and wanted to pass along my paper and the source code. I used Normalized Compression Distance (NCD)...
Hi Wesley, when I referred to social learning in society I was more looking at the implication in the opposite direction: It does appear as you point out that...
Ken, I think that you're right on that. As we showed in the paper, without evolution social learning fails pretty pathetically. Evolution therefore is adding...
Just wondering if there has been any work using hyperneat in competitive coevolution. It seems like the ability to setup the substrate to represent the...
The following Pseudo-Code is suggested for improving generalization in sparse high dimensional design spaces: 1) For each dimension individually: a) Select...
Hi all, I've been doing some research into neuroscience and the functional properties of biological neural networks that might be useful for online-learning /...
Hi Oliver, that's a good discussion topic. There's a lot of disagreement and confusion over the the question of what to take from biology and what to abstract...
hey NEAT i havent felt so productive in years http://www.cnbc11.net/050912/ i have been so rewarded by this little method thats why im trying to tell other...
Hi Ken, Yes, I'm pretty sure that not all of the phenomena I listed are important; and that a good starting point in general is to assume that they are not. I ...
Hi, My work is a 3d animation environment using digital rune engine joined by the neat algorithm I would like to load the best individual and run it in a loop...
Hi Ori, There are many different implementations of NEAT. Could you please let us know which library you're using? Wesley ... Hi Ori, There are many different...
Ori, Great. If you're using the Experiment design that Colin uses in the demos, then it's pretty easy. Assuming you've created your experiment class, you just...
Thanks Wesley, We still have some issues regrading this. We would like to run ONLY the best genome in some sort of a loop. meaning we would like to run our...
I've been working on this project to research the influence of substrate topology on a learn-to-race task. However, my harddrive crashed and now I'm trying to...
Suppose that, instead of incrementing a global counter to assign historical markings to new genes, you just use a consistent hash function on the connection...
Actually, you could just match up genes during crossover by comparing their [in, out] connection tuples for value equality (avoiding problems from hash...
Sebastian Risi and I are pleased to announce the new ES-HyperNEAT Users Page, as well as new publications on the method. The ability of HyperNEAT to evolve...
Hi Thomas, essentially what you are proposing is just an alternative form of historical marking. Whether you match up the nodes or match up the connections,...
Hello! Is anyone aware of prior work done on application of RBM unsupervised learning as a preprocessing step prior to application of NEAT? Thanks, Andy...
Hello, I am trying to evolve an ANN for the simple task for moving a car like robot to a given target position in 2D. For this i am using NEAT. I followed the...
Hello Colin, I added a Novelty fitness system to the SharpNeat 2D Bipedal Walker with some interesting results. It appeared to evolve better walkers faster...
