Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Puzzled about testsuites

Author: Michael Yee

Date: 12:24:35 03/10/04

Go up one level in this thread


On March 09, 2004 at 17:21:02, Dan Andersson wrote:

> There are several problems in automated training AFAIK:
> One of those is that you have to assume that the tree search isn't
>pathological.
> Another is that when you are using test sets the played move might have been
>chosen by a probabilistic heuristic and not by thourough search. You might call
>this the style of the player.
>
>MvH Dan Andersson

That's a good point. There would certainly be some "structural error" based on
the limitations of the class of functions we decided to train. But given the
function class (which we've assumed is the same for both hand tuning or machine
tuning), my main point is that an automatic approach (attempting global
optimization of the parameters) could get the most performance out of the chosen
function class.

Michael



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.