Using cellular automata as heuristic of computer Go

Based on the essential similarity of the cellular automata (CA) and the game of Go, this paper proposes a CA-based candidate move generator (GoCA) as well as the correlative knowledge represent method. A scheme for computer Go is also propounded. In which, GoCA perform the task of learning rules from expert game records, and then as a heuristic method, it combines with the minimax search to prune volume branches from the game tree. The experimental result shows the feasibility of anchoring our hope on it to improve the competitive ability of computer Go.