On Playing Two-Person Zero-Sum Games against Nonminimax Players
George H. Burgin · IEEE Transactions on Systems Science and Cybernetics · 1969
Results on a practical method for exploiting weakness in the play of an opponent in two-person zero-sum games are reported. Finite-state machines obtained by an evolutionary process are used to determine the move against a nonminimax opponent and achieve average payoffs significantly better than the value of the game.