Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Research idea? re: weight optimization

Author: Steve Maughan

Date: 05:06:05 12/29/01

Go up one level in this thread


Tom,

>Possibly the most important part of the idea is that the searches are
>iterative, so this suggestion will not work.

No this is how neural networks train.  There is no reason why the small changes
cannot be made after all of the positions have been examined and a new iteration
begins.  In NN some train by making changes after each observation other make
the changes after looking at all observations - I think it's called epoch
learning.  The idea is to smooth out the changes.

Regards,

Steve




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.