Author: Stuart Cracraft
Date: 05:52:58 08/16/98
Go up one level in this thread
On August 16, 1998 at 08:47:23, Larry Oliver wrote: >Most programs seem to use the old standard pawn=1, knight=3, etc, method. This >was always stated in the beginner books to be a "rule of thumb" based on >practical play. Never seemed very scientific to me. Has anyone ever tried to >optimize these values using scientific methods? Don Beal used temporal difference learning to have the program self-learn its material values, published in the ICCA journal. An interesting paper. --Stuart
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.