Author: margolies,marc
Date: 22:56:39 05/27/03
Go up one level in this thread
i doubt you are a moron, mike. please forgive me for sounding high fallutin'. let's consider a a chess playing program by its parts. there's the engine-- that analyzes chess positions. the engine can be 'tweaked' and those tweaks are called 'parameters.' An example tell the engine to value "king safety' more or bishops more than knights or even to devalue outside passed pawns. Or the size of theallocation of memroy reserved for 'hashing.' In my opinion, changing these parameters, or parametric values, would constitute 'engine learning' if it happened on its own accord. But that isn't the case. another part of the paying program is the 'opening book,' which is generally written by a chess expert, not a programmer. These men are called sometimes 'book cookers.' Gm boris alterman works on junior, eg. and gm joel benjamin worked on 'deep blue.' and there are many other fine book writers who contribute here on this board. the chessbase playing interface fritz/shredder/tiger has a book learning option. the practical results of games played by your engine on your machine can then result in changes to move selections in the opening of the chess game played by your engines on yout own machine, but not others because their books have not been modified in the same way. A chess book in the form a *.ctg tree file in chessbase can have different evaluations of the quality of moves in an opening position. And we can call these evaluations the 'weight' of the move. i hope i've cleared up the screwy code words which i used in the earlier posts for you, mike. best regards, marc
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.