Newsgroups: comp.ai.genetic
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!nntp.sei.cmu.edu!news.cis.ohio-state.edu!math.ohio-state.edu!uwm.edu!news-res.gsl.net!news.gsl.net!usenet.eel.ufl.edu!tank.news.pipex.net!pipex!usenet2.news.uk.psi.net!uknet!usenet1.news.uk.psi.net!uknet!uknet!bcc.ac.uk!uu-igor-mac.open.ac.uk!user
From: monty@watson.open.ac.uk (Tony Hirst)
Subject: Classifier sys messages over {0, 1, #}
Sender: news@ucl.ac.uk (Usenet News System)
Message-ID: <monty-070896131952@uu-igor-mac.open.ac.uk>
Date: Wed, 7 Aug 1996 13:19:52 GMT
Organization: HCRL, The Open University, UK
Followup-To: comp.ai.genetic
Lines: 26

In the 2nd edition of Adaptation in Natural & Artificial Sys, and again in
the more recent Hidden Order, John Holland presents classifier systems
whose message/action portion is defined over {0, 1}. Having just got hold
of his '86 paper 'Escaping Brittleness', the same portion is defined over
{0, 1, #}, where the wildcard symbol is used to pass through the
corresponding bit from the input, matched message.

Any ideas why the change in stance? The more recent version requires 2^n
rules to match the functionality of one in 'the old style' that contains n
#'s in the action part. Surely the old style suppports generalistaion
rather better?


tony

--------------------------------------------------------------------------
 All opinions etc etc...
--------------------------------------------------------------------------
      | Tony Hirst ("Monty")          | e-mail:  A.J.Hirst@open.ac.uk
      |
       -------------------------------------------------------------------
*******
******* The 1st Online W/S on Soft Computing will be open any day now:
******* http://www.bioele.nuee.nagoya-u.ac.jp/wsc1/
*******
       -------------------------------------------------------------------

