Incremental fuzzy clustering for an adaptive backgammon game

Modern game designs rely on artificial intelligence capabilities to create exciting gaming experience. Adaptive artificial intelligence is one of the leading capabilities which enables a computer controlled opponent to improve its logic and to cover its flaws. In this paper an adaptive AI method for a backgammon game is introduced. The specific design of this game makes its flow to rely on the two most important characteristics of most games, chance and the skill of the player. The method introduced in this paper addresses both of these characteristics by using an incremental fuzzy clustering method, derived from weighted fuzzy c-means and an incremental clustering, which adapts itself to its opponent. In the end, the merit of this paper is providing a comparison of the introduced fuzzy method and its corresponding crisp method in performance.