Higher order statistics in play-out analysis
In: The Ninth Scandinavian Conference on Artificial Intelligence (SCAI 2006), 25-27 Oct 2006, Espoo, Finland.
Playing out the game from the current state to the end many times randomly, provides statistics that can be used for selecting the best move. This play-out analysis has proved to work well in games such as Backgammon, Bridge, and (miniature) Go. This paper introduces a method that selects relevant patterns ofmoves to collect higher order statistics. Play-out analysis avoids the horizon effect of regular game-tree search. The proposed method is especially effective when the game can be decomposed into a number of subgames. Preliminary experiments on the board games of Hex and Y are reported.