Hi Jeff, Thought provoking question. Here's my gut response: Seems to me that a hidden layer is a possibility (probability) space for hidden nodes: n^2 that...
Hello Jeff, I don't think that such assumptions about layered topology are the way to achieve this. There are two fundamental things associated with any...
Peter, these are interesting thoughts and similar to the way I think about it. When I think about evolving the substrate, I usually think of a distribution of...
Hi Ken, I am happy that you find my thoughts interesting. I am trying to find a solution for this but not a constrained one like most. It has to be as general...
Hello Ken- I am glad you decided to continue the conversation. I fully agree that these are deep and difficult issues. I am also not sure how to solve them, or...
This is an exhilarating conversation. I really like the ideas that have been expressed. I agree that, in the spirit of complexification, we would want to start...
Brainstorming... and what about an evolution of a substrate in a fractal way? :P Instead of thinking to evolve the "main"substrate like neat, we could do that...
Hi all, My colleague Faustino Gomez has an open PhD position that might be of ... We are seeking an outstanding PhD student in the field of evolutionary...
Jeff, I think the issue of the right domain for evolving the substrate is very tricky. The risk is that in some domains there is little to gain by evolving...
Luca, that could be interesting and makes sense but it moves a little bit more towards more traditional indirect encodings with explicit hierarchies. I'm...
One domain I suggest is Alife. Imagine 2D creatures, whose bodies are directly derived from spatial CPPN's output and converted to mass- spring models. Now...
Joel Lehman and I are excited to announce our new publication to appear in the Eleventh International Conference on Articifial Life (ALIFE XI), called...
Hi Joal and Ken, This papers and the results presented there really turned my understandings for EC upside down. This kind of search is really innovative...
Peter, Most interesting that you bring up the concept of distance at this time. I have been reviewing various metrics of distance and weight in graphs and ...
Hi Peter, I am glad that you found this paper thought-provoking, while we were writing it my understanding of EC went through some large changes as well! You...
... how much ... of seeing ... Ken, I have to admit that although I was at UT Austin I was not very aware of the work of these two researchers. I do remember...
Peter, just to add on to what Joel said, as Joel mentions, we've been further investigating some of the issues that you raise. For example, it appears likely...
I've read Ken's latest paper on novelty search and its a fascinating new approach. For the past couple of months I've been working on virtual creature...
Hi Peter, that's a good question and of course learning to control a virtual creature with novelty search is a complicated issue (and an appealing idea as...
Thanks for your quick response Ken! You've given me a lot to think about, I just have to ponder the specifics of how to compare behaviours in an objective way...
Hi Peter, I cannot add more to what Ken said, this is an entirely new field and a cutting edge of research but I suggest you take a look at my NEVH project as...
Peter C's message brings up another issue that might be useful to Peter B: For ambulation, it might be more effective to use leaky integrator neurons than the...
Hi Ken, when you mentioned that Tyler used no inputs but just internal pattern generation, it reminded me about a question I wanted to ask you a few days ago. ...
Peter, there is no indication that the maze solvers produced by novelty search are any less "smart" than those produced by fitness-based search. As we mention...
But didn't the agent reached the goal just because the goal is very different from the start position? It's like the objective function was embedded in the...
Rafael, What you are saying is true -- if you are interested in finding a specific behavior with novelty search, then you need to design the novelty measure so...
Hi everyone! I have some comments based on my recent work with Risto Miikkulainen here at UT Austin. I evolved controller networks for quadruped and hexapod...
My goal is to evolve a stock trader. The inputs are various technical indicators, but I like the output to indicate five states--Buy/Sell (going long),...
Well I don't understand what does HyperNEAT to do with it. Unless you believe there is some geometric regularities in those time series there is no point to...
Hi Joel, What I'm saying about the embedding is that there's a high degree of correlation between the specific chosen novelty function and an objective ...
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.
