Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Puzzled about testsuites

Author: Dan Andersson

Date: 13:01:27 03/10/04

Go up one level in this thread


 Automated learning is a fun field of research. The problem must almost always
be operationalized/categorized/limited in some way.
 Global optimization is always possible, if not always feasible, inside a given
framework. The following is an argument of practicality and effifacy:
 I.e. You can always train evaluation weights inside a fixed search algorithm.
It might possibly do progress if you train eval weights as well as extensions at
the same time. But it doesn't follow that such training will progress faster
that training extensions and eval terms separately. And then combining those two
optimizations. The same might be true for categories of evaluation terms
depending on the choice of those terms.

MvH Dan Andersson



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.