Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: about Learning function.

Author: Dann Corbit

Date: 16:14:43 02/27/06

Go up one level in this thread


On February 27, 2006 at 18:25:43, Tord Romstad wrote:

>On February 27, 2006 at 17:36:24, Robson V CorrĂȘa wrote:
>
>>On February 27, 2006 at 16:50:19, Jouni Uski wrote:
>>
>>>Interestingly the best 2 engines are WITHOUT learning: Rybka + Fruit.
>>>
>>>Jouni
>>
>>
>>Yes, that's the point. many top engines without a learnig fuction. It's makes me
>>think about the real funcion or importance this function.
>
>The importance for playing strength is almost zero, I think -- at least
>with the current state of the art in position learning.  It can be quite neat
>during interactive analysis, though.

Some engines are tuned to play winning openings against a particular opponent
again and again.  You can imagine how it might impact (for instance) SSDF
results.

It clearly had a large outcome on Enrico Irazoqui's Cadaques tournament a few
years back.

The engine would probe different openings and when it found a winning line it
would play it again and again unless/until it started to lose.






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.