Pheromone-Based Pre-processing to Improve Move Generation in Board Games

Abstract

We propose a method to improve generating moves in board games. The basic idea comes from swarm intelligence. The board game program runs Pheromone-based Pre-processing algorithm (PP) to construct two heuristic game trees apart from the game search tree, and learns from winners’ moves by laying pheromone trails on heuristic game trees before game starts. Then the program picks moves on the basis of their pheromone quantity to start game tree search when playing games. The results of our experiments show that the method is efficient and promising.