Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Value of Book Learning ??

Author: Uri Blass

Date: 12:14:12 03/25/02

Go up one level in this thread


On March 25, 2002 at 15:03:25, James T. Walker wrote:

>On March 25, 2002 at 14:57:47, Dann Corbit wrote:
>
>>On March 25, 2002 at 14:53:35, José Carlos wrote:
>>
>>>On March 25, 2002 at 13:35:33, James T. Walker wrote:
>>>
>>>>I just did a quick test to see if there is any gain through book learning.  I
>>>>loaded One computer with Fritz 7 and let it learn from the 3 databases I have
>>>>(more than 1700 games played by Fritz 7.  In the other computer (both AMD 1.4G)
>>>>I cleared the book learning in Fritz 7 and played 100 games at G/1 minute (for
>>>>quick results of course).  The final score:  Fritz without previous book
>>>>learning won by 52-48.
>>>>Comments?/Conclusions?/Insults?
>>>>Jim
>>>
>>>  I don't know how Fritz 7 book learning works, but I'm certain Averno's book
>>>learning has improved results in my private tests for about 50 ELO in about 4000
>>>test games.
>>
>>I suspect that book learning is only valid for the time control at which it is
>>applied.  For example, if I ran a million games at game in one minute with
>>learning turned on, then learning will be valuable for games at that time
>>control or faster.  If I tried to use the learning file at 40 moves in 2 hours,
>>I think it would be a big mistake.  However, the learning file from one million
>>games at 40/2 would still be good for faster time controls.
>>
>>Just a guess, I have not actually examined any data.
>
>I was thinking along the same lines when I started the test.  I was just in a
>hurry to get some quick data so I ran the G/1min test.  None of the previous
>book learning was done at G/1min.

It is possible that book learning at different time control simply
does not change much.

The same program that won the at slow time control may lose at faster time
control because the opponent respond in a different way that is not a mistake.

In a lot of the cases that programs change their opinion they do it between 2
moves that have almost the same score.

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.