Predicting the opponent´s action using the k-nearest neighbor algorithm and a substring tree structure

This paper proposes a technique for predicting the opponent´s next action accurately by combining the k-nearest neighbor algorithm and a substring tree structure. Furthermore, our AI, employing the proposed technique, is enhanced by the implementation of Boltzmann equation control. The proposed method and the enhanced AI are evaluated on a fighting game AI competition platform, and their results attest to their effectiveness.