... Nothing major, I don't even remember what it was ;) Oh, now I recall, I had to change the way I find a innovation because I used to keep innovation in an...
... I was ... because I ... finding ... n). ... expensive, so ... in O ... the ... Do you keep the list of innovations around forever or do you delete it after...
... I'm sorry, I was referring to genes (they're identified by their innovation id). Genes (both nodes and connections) are sorted by their innovation id. When...
The steeper the hill you're climbing, the more important elitism is. It's an intuitive concept in search: you don't want to forget the highest point you've...
Since the documentation asked for those considering an experiment for NEAT to let the author know, I figured I'd write in... I've read several papers that...
... for NEAT ... for image ... improvement ... which ... photos and to ... conjunction with ... engine. ... may have ... if I get ... By all means, and other...
Following up on a previous discussion on solving Tic-Tac-Toe, I wanted to open up to the group the question of the rule base for optimal play. I tried to find...
Determining if the rule set is optimal or not is doable, though perhaps time consuming. There are 10 possible starting conditions: rule player goes first, or...
... organize my ... John, thanks for the nice explanation of image enlargment, something I knew little about! One thing that may be a challenge in this domain...
... The bigger problem for the moment is extracting the training data within this lifetime. :) My first attempt was only processing at a rate of about 25...
... What is your source for a training set? ... Just to try to clarify...your network topology is a 4x16 fully-connected feedforward network? Is that right?...
... I have almost 3000 pictures that I have taken with my digital camera. I don't want to enlarge all of them, but since I want to be able to enlarge any one...
Hi John, Thanks for elaborating...sounds like a fun and interesting domain. ... What was your approach in gathering a sample of test photos? Were you trying...
... Actually, I'm using 'favorites first.' That is, if I only have enough disk space for 100 photos, I'm going to use the ones I plan to enlarge first. If ...
A number of people like John have recently brought up letting NEAT decide for itself which inputs it should use. (As an example the below quote is from John) ...
Some of you may have read our previous Tech Report, "Competitive Coevolution through Evolutionary Complexification." That paper has now been replaced with a...
... Cool...thanks for the update, Ken. I'm interested to read this updated paper. I had a question about how you implemented your robot simulation. Did you do...
Hi Derek, I wrote my robot control code off of previously existing Khepera simulator software. So the core of the code is from that, although it's been much...
Hi Ken, If it's not to much trouble I'd like to ask you a few basic questions about specific implementation details in NEAT. Just to explain where I'm coming...
... Ah, okay...cool. ... TeamBots does have a "graphics-off" mode, so it does look like a good start. I also believe that the SoccerBots extension allows you...
As I had posted earlier, NEAT does great in robot control domain. I had tried it out on a from scratch simulation, and had nice results (atleast for gathering,...
I was reading Ken's updated paper on Competitive Coevolution through Evolutionary Complexification, and I came across the Evolution of Complexity section ...
Hi Derek, Ken ... [...] All good questions. I have to admit that I've never fully convinced myself that a population's complexity (within NEAT) wouldn't...
Hi Colin, absolutely feel free to ask questions. I think a C# implementation of NEAT would be great, and also I agree doing you own implementation will be...
Derek, good question and I definitely understand where you're coming from. The key point that is perhaps not clear is that the lines in the middle are *not*...
... Hmm...why not? ... Yeah. :) Philip and I have been working on our implementation for quite a while now. We've learned a lot in the meantime, but it'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.
