Newsgroups: comp.ai.neural-nets
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!news.duke.edu!convex!cs.utexas.edu!news.sprintlink.net!EU.net!CERN.ch!news
From: Owen.RANSEN@cern.ch (Owen RANSEN)
Subject: Re: Chess and Neural nets
X-Nntp-Posting-Host: misf120.cern.ch
Message-ID: <D66wC7.9nL@news.cern.ch>
Sender: news@news.cern.ch (USENET News System)
Organization: CERN - European Organization for Nuclear Research
X-Newsreader: WinVN 0.93.11
References: <3kl3lb$27a@cantua.canterbury.ac.nz> <D5sKDE.Eo7@news.cern.ch> <3koj2j$elk@bradley.bradley.edu>
Mime-Version: 1.0
Distribution: usa
Date: Wed, 29 Mar 1995 06:50:30 GMT
Lines: 32

In article <3koj2j$elk@bradley.bradley.edu>, pwh@bradley.bradley.edu 
says...
>
>Owen.RANSEN@cern.ch (Owen RANSEN) writes:
>>Basically backgammon is great for neural nets, and chess isn't.
>
>
>"why"? :-)
>
>The variable values of the pieces, as opposed to backgammon?
>-- 
>Pete Hartman                   Bradley University       
pwh@bradley.bradley.edu
>                                 "I'm trying!"
In my original mail I said that I was quoting from a paper by Tesauro
on a backgammon playing neural net, so I'm not the person to really
reply why.

From what I understand tho:

    1) Backgammon has ideal inputs to a nerual net, 26 integer values on
       the 24 + 2 points of the board. Each integer value is the number
       of pieces on that point. I'm not sure waht the vector of inputs
       for a chess board layout would be. It is certainly not as 
       obvious as backgammon
    2) Backgammon "naturally" randomly explores the possible evolutions
       of the game, because the dice throws are random. Unless you        
       introduce an "artificial" element into the chess player chess does
       not have a natural random component.
 


