Newsgroups: comp.ai.genetic
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!oitnews.harvard.edu!purdue!haven.umd.edu!news.umbc.edu!eff!news.duke.edu!news.mathworks.com!newsfeed.internetmci.com!in2.uu.net!pipeline!psinntp!psinntp!psinntp!megatest!news
From: Dave Jones <djones>
Subject: Speciation
Content-Type: text/plain; charset=us-ascii
Message-ID: <DJqAyw.1As@Megatest.COM>
Sender: news@Megatest.COM (News Admin)
Nntp-Posting-Host: pluto
Content-Transfer-Encoding: 7bit
Organization: Megatest Corporation
References: <4a7kf9$4db@hippo.shef.ac.uk> <4ac1j0$12v8@majestix.uni-muenster.de>
Mime-Version: 1.0
Date: Sun, 17 Dec 1995 11:37:42 GMT
X-Mailer: Mozilla 1.1N (X11; I; SunOS 5.4 sun4m)
X-Url: news:4ac1j0$12v8@majestix.uni-muenster.de
Lines: 28

I'm wondering if anyone has investigated the idea of
species-recognition in genetic algorithms. I'm thinking
of a "match-maker" physics that helps organisms pair with
suitable mates for reproduction.

In the wild, organisms eventually develop elaborate mechanisms
for identifying potential mates, and THE number-one criterion is,
the mate must be of the same species. Mating calls, dances,
displays, sniffing, and what-not. Useful stuff, all this,
because he mating of different species, by definition, produces
uncompetative offspring, either still-born or incapable of reproducing.

It seems to me that genetic programs may very often produce
distinct species, only to see all but one quickly become
extinct. The dominant species finds compatible, fecund mates, while
the others waste reproductive generations attempting to
mate with the incompatible, but more numerous, dominant
species, and become extinct through bad-choice attrition.
Genetic programs generally don't give their populations
the opportunity to direct the mate-selection process, and
even if they do, they probably do not have the survival
"niches" that would allow minor species to survive long
enough to evolve effective mate-selection criteria.

Any ideas?

             Dave

