Newsgroups: rec.games.go,comp.ai.genetic,comp.ai.games
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.mathworks.com!fu-berlin.de!news.dfn.de!news.uni-augsburg.de!news
From: PitBull <schaefer@malaga.math.uni-augsburg.de>
Subject: Re: Q:Using Genetic programming to find GO-evaluation function
Content-Type: text/plain; charset=us-ascii
Message-ID: <Dn4Mzt.34K@Yuri.CC.Uni-Augsburg.DE>
To: sdsilva@wi.leidenUniv.nl
Sender: news@Yuri.CC.Uni-Augsburg.DE (The News Administrator)
Nntp-Posting-Host: jaen.math.uni-augsburg.de
Content-Transfer-Encoding: 7bit
Organization: http://malaga.math.uni-augsburg.de/schaefer/
References: <312882A7.5492@wi.leidenUniv.nl>
Mime-Version: 1.0
Date: Wed, 21 Feb 1996 13:04:39 GMT
X-Mailer: Mozilla 1.12IS (X11; I; IRIX 5.3 IP22)
X-Url: news:312882A7.5492@wi.leidenUniv.nl
Lines: 36
Xref: glinda.oz.cs.cmu.edu comp.ai.genetic:8064 comp.ai.games:3893

"S.F.da Silva" <sdsilva@wi.leidenUniv.nl> wrote:
>Hi,
>
>I'm a student in computer science and I'm currently working on a GP to
>find an evaluation function for Go.
>
>The evaluation function works as follows:
>  for each point on the board it looks at a 5x5 square on the board around
>  that point and, according to the stones present there, it places a black
>  or white stone or nothing on a new board at the coordinates of the 
>  previously mentioned point.

Yes, it may be useful to use AI techniques to get guesses for best moves.

>  After that it counts the number of black and white stones on the new board.
>  The difference between these numbers will be the evaluation value of
> the board.

That's nonsense. The value of a position on the Go Board does not depend
on stones alone1. Even if it did, it would require searching a large tree
to detect the future effects of a move.
I think you need an expert system to calculate the value of a board position.
You should use a different representation of the board. 5x5 doesn't say lots
about a position.

-- 
Peter Sch"afer

"office":
schaefer@malaga.math.uni-augsburg.de
http://wwwhoppe.math.uni-augsburg.de/schaefer

"leisure":
schaefer@mathpool.uni-augsburg.de     
http://www.mathpool.uni-augsburg.de/~schaefer

