Author: David Mitchell
Date: 03:39:34 07/07/04
Go up one level in this thread
On July 07, 2004 at 00:50:31, Gopikrishna wrote: >When I matched my engine with another one it played all the first 75% moves >similar to the previous game.Although both the games were drawn.But my engine >had a slight advantage but it went for a drawing line of 3 fold repetition :( >So how to alter from this? Maybe implementing any learning will improve this? >Any suggestions please? Thanks. What do you have as a threshold value for your program to choose either to work for the win, or to seek a draw? Also, do you use a contempt factor along with the above threshold value? Learning is to avoid blunder moves that look good in a search, where the real evaluation of the resulting position can only be determined with a good deal of work (time). Learning will not change your program's propensity to seek out a draw, even when slightly ahead. You should have an explicit variable to handle that. Dave
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.