Author: Christophe Theron
Date: 23:02:05 10/18/97
Go up one level in this thread
On October 19, 1997 at 00:19:15, Robert Hyatt wrote: [snipped] >But the problem is real, and serious. The commercial programs suffer >from >this quite seriously... to the point that the operators of many of these >programs won't play the humans that "can draw almost at will" (this in >the words of one genius operator on ICC, found in his finger notes.) :) Yes. We shall work on this later. But we have another problem: humans learn so quickly how to beat our programs, I'm wondering if we could write polymorphic (changing) evaluation functions to confuse them. It goes beyond book learning. Some say the DB team changed some terms of their evaluation during the match against GK to avoid this. It seems smart. - Christophe -
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.