... Food for thought. I experimented with a tree based genome structure last year and was intending to take a fresh look at that work once I had a working NEAT...
I've made some changes since the last time I posted a demo. The newest version is available here: http://www.vrac.iastate.edu/~streeter/avh_spring_2004.html ...
... I would agree with this except it's not just the inner structure that is affected. The *outputs* of the network will tend to all be 0 or 1. You might say...
Dear NEAT list. I would like to ask a simple question. I begin quoting ken in his main NEAT paper: "We present a novel NE method called NeuroEvolution of...
Hi German, I'll go ahead and field this one, and let Ken add any comments if need be. ... A really common problem in Neuro-evolution (and in complexifying...
Joe pretty much has it. Just to add to that, it used to be that people who designed topology-evolving algorithms would usually support their use by explaining...
I wasn't intending to put out another release just yet as I am in the middle of a code reorganisation. However I though some might like to see the various...
On Sun, 2 May 2004 23:40:54 -0500 (CDT), Joseph Reisinger ... Honestly I haven't seen much of this approach in stuff I've read. Is there a name for this type...
... use ... from ... not ... I think this is missing the point, though. This is exactly the perspective I was trying to argue we should change. The question...
Ken, I think I may be missing your point somewhere. I am interested in understanding. As such, this is how what your wrote occurred to me. ... we care if it...
... On Mon, 03 May 2004 20:27:47 -0000, Kenneth Stanley ... <snip> ... Right. Why would we care if our resultant champions have 25% unnecessary topology?...
Jim and Derek (I am responding to your posting here too), I probably should have put a bit more time into my short little explanation because obviously I...
... Well, in common case we don't know whether found solution has minimal topology or not. So we can't define what part of result net is unnecessary. This...
... think ... Okay...fair enough. ... Right...understood. ... Right...but if networks in your final generation contain redundant or unnecessary topology, then...
... While this is true, I think the important point is the opposite: just because there is a nice final topology does not mean that you didn't waste a lot of...
SharpNEAT looks like it's starting to work really well, and the simulations are great. What is the fitness function being used in the no wiggle pole...
... Yep, cheers. Starting to get some nice results now. Actually I was just looking at Mattias' pole balancing demo - on the DelphiNEAT web site Mattias...
how do you differentiate between similar types of information with NEAT? I understand that the input should represent intensity, but what if I want to add more...
... Hi Chad, I would keep the monochromatic input and attach 3 new color signals. That way previously evolved genomes can be used (with the new inputs added)...
so let me try to summarize what you said. If I find a network that achieves some minimum fitness, adding new nodes and continuing the search would be the best...
Hi Chad, this idea of freezing certain preevolved portions of a network and letting it build on those is an interestng and viable research direction. It is...
Hi Chad, ... In some circumstances, yes. If we take a problem that requires a network with color inputs as an example, we might not find a good solution if the...
Perhaps it's possible to apply delta coding as explained in ESP to search in the vecinity of the solution space rather than freeze the resulting network (could...
What's the easiest way to do speciation visualization without writing a lot of code? I'm using Ken's original source, but in a cross-platform way... so I ...
This is a tech question, no theory: I'm using mat bucklands windows neat package, and the population from genotype from text file function is only a stub. I...
On Wed, 5 May 2004 22:03:16 -0700 (PDT), Tyler Streeter ... I've been meaning to ask Ken more details about how he produced the speciation graphs in his...
... Correct me if I am wrong, but it seems to me that if you include a complexity penalty in the fitness function, you create an evolutionary pressure towards...
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.
