Author: Kurt Utzinger
Date: 12:35:33 08/03/05
Go up one level in this thread
On August 03, 2005 at 14:30:13, Peter Skinner wrote:
>On August 03, 2005 at 14:19:18, Kurt Utzinger wrote:
>
>>On August 03, 2005 at 14:04:22, Peter Skinner wrote:
>>
>>>Hi Dan,
>>>
>>>Since Fruit doesn't have the ability to learn, would an interface like Arena or
>>>Chesspartner handle it for it?
>>>
>>>Peter
>>
>> Hi Peter
>> Why do we pay so much interest in the ability of learning.
>> Have we ever seen/heard/read about a significant difference
>> in Elo between learning on/off? In my opinion this feature is
>> much overestimated.
>> Kurt
>
>So if an engine can lose the same way over and over again it is not important
>that is learn not to fall into the same traps?
Show me an engine which has lost the same way over
and over again. This happens very seldom due to the
many different book lines used.
Kurt
>
>Learning is very valuable just for that point alone, not withstanding the
>benefits of positional learning as well.
>
>I believe it to be _very_ important that an engine learn what not to do, and
>even more, what it did right to get the win.
Right, but I dont's see the point if you can't
measure the advantage of this feature. The few
games (draw instead loss / win instead draw)
do almost have no influence on the score of an
engine over some hundred/thousand games.
Kurt
>
>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.