Author: Uri Blass
Date: 12:52:15 11/10/00
Go up one level in this thread
On November 10, 2000 at 14:54:24, Uri Blass wrote: <snipped> >There is a problem what to do against new opponents and in this case it is >possible to try both engines in the first 12 games and decide based on the >results to learn which program to use in the next 28 games. > >I do not know if 12 is the optimal number. The optimal number should not be a fixed number and it should be based on previous results. For example if the results of both engines are the same after 12 games it is logical to continue to try them. The problem with the idea of choosing the engine based on the opponent is the fact that I cannot use the information to predict results against an unknown opponent and there are cases when you have to play against unknown opponents in tournaments. Uri
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.