Newsgroups: comp.ai.genetic
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!howland.reston.ans.net!Germany.EU.net!EU.net!julienas!news.fnet.fr!efrei!vaissier
From: vaissier@efrei.fr (Amaury de la Vaissiere)
Subject: Re: Dynamic changing of fit. func. - Evolvable Fitness Formula
Message-ID: <1994Nov25.95401.21588@efrei.fr>
Organization: EFREI
Approved: usenet@efrei.fr
X-Newsreader: TIN [version 1.2 PL2]
References: <3av5o7$75p@belfort.daimi.aau.dk>
Date: Fri, 25 Nov 1994 09:54:01 GMT
Lines: 77



Henrik Hautop Lund <hhl@daimi.aau.dk> wrote:

Excuse me for criticising, but your explanations are not very clear to me.
Your ideas migth be very interesting but I am not shure that we are talking
about the same thing.

>The fitness formula (ff) represents
>suitability of an individual to it's environment.

Not exactly (to me). I would say that it represents a _way to estimate_
the suitability of an individual to it's environment.






For example, (to fix the concept) lets take an individual trying to
approximate a function f(X) of a real variable X. For a given Xi, the 
approximation of that invidual is a(Xi).

Then the fitness formulla for that individual might be:

fitness = 1 / Sum ( | f(Xi) - a(Xi) | )

              for i=0 to Max

Max depends on the precision of the fitness wanted.


Now, in your GA, how would you implement an adaptive ff for that example
(or an other example, as simple, so that every one can understand your
technique)?






I admit that in real life the ff is not explicit. There is no quantification
of "the ability to survive" of an organism. It is the ultimate goal,
but there are many "sub"-goals different to each individual and specie.

Furthermore, in very complex environements it is generally difficult
to imagine an appropriate ff. Is is a nice idea to divide this big goal into 
smaller ones easier to achieve. It is also a good idea to introduce genes  
in the individual so that it can "build" and optimise its own small goals
(for example: retreive food, protect itself, ...).

But I really can't see how you would do this partitionning.

What troubles me also is that you don't explain clearly how the individuals, 
in your GA, will be selected.






I understand that the answer to my questions are certainly in the two
articles you are refering to. I will look at them when I will have an 
access to ftp, but could you give us a short answer for the moment.

Thank you.




    ___
  /|   |\	
 / |___| \       I juge that the existence's purpose is the most urgent
| /     \/|   question. How can we answer it?                 Albert Camus
|/\ ___ / |
 \ |   | / Amaury de la Vaissiere <vaissier@efrei.fr> GCS/E/MU/S -d+(?) p---@
  \|___|/ c++ lu++(+) e++(*) m*(-) s+/- n-(--) h*(+) f++(?*) !g w+++ t+++ r !y
