Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: automated book generation in chess/checkers

Author: martin fierz

Date: 13:31:11 12/13/01

Go up one level in this thread


On December 13, 2001 at 13:32:08, Dan Andersson wrote:

>So thats dropout expansion. It seems a lot like some of the ideas bandied around
>in the mid eighties. It goes to show that each generation must learn some old
>tricks anew. I have seen this in action/strategy games programming too. Old and
>well known tricks re-discovered and presented as new, as they are to the new hip
>crowd. Only shows that common knowledge should be codified and written down, no
>matter how well known. Sometimes it seems like the comp industry is brain wiped
>every five years.
>
>MvH Dan Andersson

i don't know too much about it, tom also uses some additional heuristics i didnt
describe, like if the conspiracy number of a node is small it's expansion
probability rises too.
what i do know is that it works very well for checkers, the book we produce is
good. i don't know how many computer-generated opening books there are which
compare well with human theory for that game, i don't think there are many :-)

regards
  martin



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.