Duplicated parameters inside evolved neurons 
Submitted by ianww on Sat, 2009-08-08 08:37

I'm a new user, and have been evolving a creature's neurons. After evolution, many neurons have duplicated parameters, for example:

[N, fo:0.150869, in:0, si:3, -2:4.233,fo:0.04,in:0.979792,in:0.8]

Am I right in thinking that it will be the last (reading left to right) setting that applies and that the earlier settings are redundant and can therefore be deleted without harm (I've tried this and it seems to be the case).

Thanks in advance for any advice that can be offered.

Cheers

Ian

(Thanks for the great software Maciej - it's something I've been wishing for for years - I don't know how I failed to come across it before!)

Forums: 
Users
Maciej Komosinski's picture

Submitted by Maciej Komosinski on Sat, 2009-08-08 10:52

Yes, you are right. The f1 genotype is first converted to f0 (you can see the f0 genotype in "Conversions"). Then, based on the f0 genotype, the creature model is built. Each new neuron gets default values of parameters, and then, while the genotype is interpreted, new values are set from left to right, so the last ones win.

Related papers are
http://www.framsticks.com/bib/Komosinski-and-Ulatowski-2009
http://www.framsticks.com/files/common/ComparisonGeneticEncodings3DAgent...
http://www.framsticks.com/files/common/GeneticMappingsInArtificialGenome...

We are glad you like Framsticks :-) The possibilities of performing experiments are only limited by imagination!

Submitted by ianww on Sun, 2009-08-09 10:36

Thanks for being so responsive Maciej!

Cheers

Ian