... are ... is ... their ... that ... point ... Could you be kind enough to explain how speciation parameters work in NEAT? What if I change these parameters?...
Well basically you need to understand how two genotypes (NNs) are compared so that it is determined whether they are in the same species or not. The total...
Thanks for your kindness, but, well, I actually know what these parameters mean, but *don't* know how to decide the parameter values in a specific problem. I...
Hm. It is hard to tell you why, but.. maybe the population size is too small? If the population size is too small and there are many species, there is not...
Well, do I have to use a GA to determine the appropriate parameters? I think the network stucture is noly related to output, no matter the input, is that...
Using a GA to determine the best parameters will require a lot of CPU time. Every individual running its own evolution.. Nah, this is too computationally...
Maybe the thing to do is to run a study of what happens for a range of different parameters for a few different test problems, then write up the results as a...
Beware: evolution is very bad at choosing good settings for evolution. That's in addition to all of the extra computation time. Cf. Clune, J., Goings, S.,...
... This is a rather broad statement, and, paper or not, I don't buy it. I believe a Differential Evolution would work *very* well to choose NEAT parameters. ...
There may be some mileage in using multi-objective evolution where instead of using parameters, the things they change are factored in as extra dimensions to...
I recognize it is a broad statement, but after doing research on this subject for a few years it is what I have concluded. I am currently working on a...
The below posts dated 02/03/07, 12/02/06, 10/18/06 and 10/12/06, may be applicable. ##################### Post Dated: 02/03/07: ##################### The...
Your point could be interpreted to be substantiated, if you were to start your parameter evolution from scratch. But, if you use the starting genome as the...
Are you saying that because you have tried it, or because you think that starting with a good set of parameters on the genome it would get better (or stay as...
Based upon experience (i.e. Supervisor NEAT). And also, yes, it is profoundly more effective to start with your best "knowledge" to date, whether supervisor...
It is very cool that you have supervisor neat going. I could not find the top-level result of the proposed experiment in there. Do the runs with an evolving...
a) The actual experiment is addressed (i.e. not proposed). b) Yes, on all problems addressed to date. c) Yes, digest the meaning/implications in the data...
This brief tutorial is intended to convey a means by which a starting genome may be created to correspond to a desired starting point. This usually represents...
Well what about just saving the best genome from previous runs? Incremental evolution? I know this introduces potentially unnecessary complexity but that...
Has there been any work to date on implementing recessive genes in a NEAT genome? I am interested in embedding additional information in a genome without...
Andy, Not that I know of...maybe Ken knows of NEAT implementations of recessive genes. At one point with ANJI, we discussed going down this route. Since genes...
Derek, What prompts my question is an experiment which employs a local gradient search, utilizing the organism output vector as the start point. The result is...
There has been work with GAs that have diploid genomes but no work specifically with NEAT that I know of. Here's a general GA paper on the subject: ...
... Andy, Not NEAT related, but my copy of Melanie Mitchells intro to GAs refers to work by Collins and Jeferson in one context , then adds in that a different...
Hi Stanley and everyone in the list, I would like to know if there is any kind of implementation and experiments done using the NEAT with self-organizing maps ...
Investigation of redundant organisms in NEAT using standard param values has shown an approximate 10 percent redundancy factor as follows: Population Size: 100...
Hi, It would be interesting to see the percentage figure for other population sizes, # of inputs/outputs etc. to see if the problem hovers around the 10% ...
Andy, thanks for pointing this issue out. Those are interesting statistics. It is indeed best to avoid reevaluating an organism that is the same as one that...
Cicero, I don't know of any published work in which NEAT was combined with a SOM. However, in my classes students have used NEAT to evolve the function that...
