Potential Applications of Opponent-Model Search - Part 2: Risks and Strategies

H IIDA, JWHM UITERWIJK, IS HERSCHBERG, HJ VANDENHERIK

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

An opponent is modelled by an assumed knowledge of his evaluation of positions in a game. The opponent may be fallible. Acting on one's model of the opponent, we discuss the risks of doing so, finding a class of riskless procedures. With respect to possible gains and losses, it is shown that these may be combined into distinct strategies, four of which are exhibited and characterized by their presumed balance among them.
Original languageEnglish
Pages (from-to)10-14
JournalICCA Journal
Volume17
Issue number1
DOIs
Publication statusPublished - Mar 1994

Cite this