Newsgroups: comp.ai.genetic
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!newsfeed.pitt.edu!gatech!newsfeed.internetmci.com!uwm.edu!spool.mu.edu!munnari.OZ.AU!news.mel.connect.com.au!harbinger.cc.monash.edu.au!news.rmit.EDU.AU!matilda.vut.edu.au!rhh
From: rhh@matilda.vut.edu.au (Robert Hinterding)
Subject: Re: Mutation rate as a gene
Message-ID: <DMuqHL.CM0@matilda.vut.edu.au>
Organization: Victoria University of Technology
References: <4g0lmc$ji6@dub-news-svc-1.compuserve.com>
Date: Fri, 16 Feb 1996 04:44:09 GMT
Lines: 42

103120.75@compuserve.com (Patrick Murphy) writes:

>It would be easy to set the mutation rate itself as something that
>can mutate.  I've heard various strategies that set the mutation
>rate of the organism based on outside factors such as how converged
>the population is.  My question is what happens if the mutation
>rate is simply another gene in the chromosome that can mutate.  My
>guess would be that over the short term chromosomes would evolve
>extremely low mutation rates, but perhaps would be selected out in
>the long run.

What you are talking about is called "self-adaption" of mutation.  It has
been used in Evolutionary Strategies, Evolutionary Programming and
Genetic Algorithms.  When self-adaption of mutation is used, gaussian
mutation rather than bit-flip mutation is generally used and the extra
gene(s) represent the standard deviation to the gaussion noise used.
So the "strength" of mutation rather than the mutation rate is controlled.
This method is quite successful and gives improved results.
References for papers on this are:

Hinterding, R., Gaussian Mutation and Self-adaption for Numeric Genetic 
Algorithms",Proc. Intrl. conf on Evolutionary Computation 1995, IEEE press,
PP 384-389.

Baeck, T, and Schwefel, H-P., An Overview of Evolutionary Algorithms, In:
Evolutionary Computation, Vol 1 No. 1,1993

Saravanan, N., and Fogel, D.B., Learning Strategy Paremeters in Evolutionary
Programming: An Empirical Study, In: Proceedings of the 3rd Annual Conf
on Evolutionary Programming, 1994

You can pickup copies of my papers on the subject which contain further
refs via my home page.

Hope that helps
Robert Hinterding

-- 
Robert Hinterding
VICTORIA UNIVERSITY OF TECHNOLOGY       Email: rhh@matilda.vut.edu.au 
P.O. Box 14428, Melb Mail Centre        Fax:   +61 3 688 4050
AUSTRALIA 3000                          Phone: +61 3 688 4686              
