Comparative Analysis of Evaluation Functions for Minimax with Alpha-Beta Pruning Applied to the Jaguar Game


Jaguar Game is a board game from the Brazilian indigenous cultural tradition. It is played on an asymmetric board with 31 positions in which a jaguar duels against fourteen dogs and where the adversaries have different goals. In order to present preliminary results regarding the use of Artificial Intelligence techniques in this domain, the game tree complexity was estimated by a brute force approach with prunning and the resulting complexity was verified as being superior than the Checkers game. We also proposed and evaluated utility functions for the Minimax algorithm with Alpha-Beta pruning optimization. Results from simulations emphasize the challenges in designing utility functions for the dogs players.

In: XVI Encontro Nacional de Inteligência Artificial e Computacional