Newsgroups: comp.ai.games
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!oitnews.harvard.edu!purdue!lerc.nasa.gov!magnus.acs.ohio-state.edu!math.ohio-state.edu!howland.reston.ans.net!newsfeed.internetmci.com!in1.uu.net!news00.sunet.se!sunic!mn6.swip.net!plug.news.pipex.net!pipex!dish.news.pipex.net!pipex!bt!btnet!dispatch.news.demon.net!demon!sunsite.doc.ic.ac.uk!nntp0.brunel.ac.uk!strath-cs!str-ccsun!udcf.gla.ac.uk!newsfeed.ed.ac.uk!ainews!andrewt
From: andrewt@aisb.ed.ac.uk (Andrew Tuson)
Subject: Re: Technological stuff
Message-ID: <DHGpy4.932@aisb.ed.ac.uk>
Sender: news@aisb.ed.ac.uk (Network News Administrator)
Organization: Dept AI, Edinburgh University, Scotland
References: <htAGxup.predictor@delphi.com> <475kf6$1k7@news.xmission.com>
Date: Fri, 3 Nov 1995 10:18:03 GMT
Lines: 25

In article <475kf6$1k7@news.xmission.com> jolsen@allencomm.com (John Olsen) writes:

>Couldn't you technically group many SA algorithms into the Evolutionary
>category?  You test a set of permutations, and choose the best result,
>making it a gene pool of one subject with mutation to produce offspring.

The answer is yes. GAs, SA and Tabu Search (yet another optimatisation 
technique), can all be viewed as forms of neighbourhood search. All have 
proven useful for many real-world problems.

A hot topic of research at the moment is to identify for what problems are the
techniques above most suitable. For instance GAs seem a good bet if you: need
to run the algorithm on a parallel computer; need multiple alternative
solutions; are attempting Pareto-optimisation problems; and if the structure
of the problem makes crossover a particularly effective operator.

Anyway, I can give more details on this if you like (it's the general subject
area of my PhD).

-- 
Andrew Tuson (andrewt@aisb.ed.ac.uk)

Department of Artificial Intelligence, University of Edinburgh, Scotland, U.K.
An expert is a person who avoids the small errors while sweeping on to the
grand fallacy..........:-)
