Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: chess learning vs better search

Author: Matthias Gemuh

Date: 07:28:52 03/04/06

Go up one level in this thread


On March 04, 2006 at 09:43:31, walt irvin wrote:

>there has been a huge effert in making chess engines great at search
>rybka,shredder,fritz,hiarcs ect ect,,,,,how much effert has there been on a good
>learning feature????? i maintain a chess program should not have to analyze the
>same position x2 once the position has a history,,,,,as a very radical example
>think of this   some one writes a chess program and calls it darwin ,,the way it
>makes its move is first it checks its position database if the position is there
>it makes the next winning move or drawing move otherwise it makes a legal move
>at randome,,,,,then you play this thing auto vs all the best programs with
>various books 50,000 games,,,, if nothing else he would remember every single
>position and the exact move to make from all those games and would now play the
>moves those programs played ,,,,,,so now u would have a program with no
>evaluation what so ever competing on a equal bases vs the best programs imagine
>if that program that learned like that were shredder or fritz ???????????? just
>a thought that learning has not been developed like searches have,,,,,,,,,
>
>walter irvin



To go this way, every engine would need TeraBytes of disk space
and same amount of RAM (i.e. memory).
I think the only part of engine development that has been neglected is
automatic tuning of evaluation parameters.

Matthias.





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.