Author: GuyHaworth
Date: 02:26:25 06/25/02
Go up one level in this thread
Generically, I guess 'learning' is a process enabling one to approach the 'same' situation the next time with a different 'mindset', hopefully a better one. As Rob Hyatt has said, the opening book mechanism needs to be a learning one benefiting from past experience. In the extreme, one does not want to lose the same game twice. Rob wrote about "Book Learning - a Methodology to Tune an Opening Book Automatically" in the ICCA Journal, Vol. 22, No. 1 (March, 1999). Equally, if armed with endgame tables, one could assess the competence of the opponent from their moves and finesse one's choices of endgame moves using that information. Jansen wrote about this in Vols. 15 and 16 of the ICCA J ... and I've just tabled a 'model of fallible endgame play'. J Fu(e)rnkranz wrote about "Machine Learning in Computer Chess: the next generation" in ICCA J Vol. 19, No. 3 (Sept 1996) ... and his work is on the web: http://www.ai.univie.ac.at/~juffi/lig/lig.html Hope this helps. g
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.