Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: SSDF. Gandalfs LEARNING is destoyed when aborting!!

Author: Uri Blass

Date: 11:25:33 10/18/00

Go up one level in this thread


On October 18, 2000 at 14:09:00, Mike S. wrote:

>This kind of problem could have been avoided since the start of the list, if
>they would have done it like this (like I wrote once or twice here too):
>
>1. switch off all (permanent) learning
>2. remove all doubles (!!) from the results

In this case you may get a match of only 2 games when both sides always repeat.

>
>By this, they would test "out of the box" versions. What's the use of testing
>software with (an increasing mass of) individual learning data, which any other
>user *does not have*? Sooner or later, they are testing something completely
>different from the software one can buy.

I am not sure about it.
Suppose a program learns to have better evaluation function from games(I do not
know about programs who do it but I believe that it is possible).

Is it logical to switch off this learning?

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.