Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: chess learning vs better search

Author: walt irvin

Date: 07:52:52 03/04/06

Go up one level in this thread


On March 04, 2006 at 09:56:10, Charles Roberson wrote:

>
>     There is a lot of effort in machine learning applied to chess
>      and to other areas.
>
>     What you describe is one of the worst ways to do it.
>      The random part is the killer to your thought.
>      The number of total positions in all possible chess matches is far
>      greater than any database can handle.

that may be true but im sure that 50,000 or 60,000 games is not that hard to
do,,,,besides if u used a strong program instead of that random one i talked
about you could store positions up until say the score of + or - 3.5 should also
eliminate alot of positions,,,,,

on the more exoctic side u could use pattern recognition ???? im sure that would
not be too easy though ????


but i dont think u need   e2-e4 and mate in 127 heehhehehe just need to get a
substantial advantage out of the openning,,,,,,,a decent program dont need a
huge advantage to win a game,,,



walter irvin
>            Just look at the work on endgame table bases -- they create a
>                database of all possible board positions and moves for 3,4,5
>                and 6 pieces on the board. They are quite large and take a
>                huge amount of time to generate.
>     Better to do legal moves instead of random ones. Also, better to not even
>     try moves that the search indicates are bad or at least very bad.
>
>      I could go on for pages on this. I suggest you do a paper search in
>     google if you want to continue your research on this. Topics could be:
>     machine learning, position learning, bayesian learning, neural networks,
>     temporal difference learning.......



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.