Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: How I can be sure? (was Rebel learning)

Author: Ed Schröder

Date: 06:55:07 06/23/98

Go up one level in this thread


>Posted by Peter Herttrich on June 23, 1998 at 06:53:21:

>Hello Ed, hello World,

>now I could also detect games, which are doubles and where
>R9 get lost.
>Surprising for me, this games are only against Fritz5.
>Of course without the 232-interface. Only the Peter-interface :-)

>Now my question, how I can be sure, that Rebel learns?
>(the menuponint is 'on', of course)

The Rebel9 learner is split into 2 parts, a part for learning against
humans and a separate part for learning against computers (for the
moment only detected by the use of auto232).

The human part of the learner is tuned to be very mild. This means you
can have the pleasure of beating Rebel9 3-4 times on the same opening
but then it's definitely the end of all the fun.

The second reason for the mild human learner is that it will avoid to
endlessly repeat opening lines Rebel9 has won. How irritating! I want
to keep my customers :)

Then there is the comp-learner. This learner is not mild but aggressive.
It will seldom replay lost openings and will strongly try to repeat won
opening lines.

- Ed -


>cheerio
>Peter




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.