A methodology for learning players' styles from game records. (28th September 2011)

• Record Type: Journal Article
• Authors: Levene, Mark; Fenner, Trevor
• Is Part Of: International journal of artificial intelligence and soft computing. Volume 2:Number 4(2011)
• Page Start: 272
• Page End: 286
• Publication Date: 2011-09-28
• Subjects: temporal difference learning -- evaluation function -- game records -- player's style -- computer Chess; Artificial intelligence -- Periodicals; Soft computing -- Periodicals; 006.305
• Languages: English
• ISSNs: 1755-4950

Abstract

We describe a preliminary investigation into learning a Chess player's style from game records. The method is based on attempting to learn features of a player's individual evaluation function using the method of temporal differences, with the aid of a conventional Chess engine architecture. Some encouraging results were obtained in learning the styles of two Chess world champions, and we report on our attempt to use the learnt styles to discriminate between the players from game records, by trying to detect who was playing white and who was playing black. We also discuss some limitations of our approach.