Abstract
In this paper we examine the evolution of adversary game strategies in an uncertain environment. Each player has a value which indicates player's ability. In the adversary game in this paper, two players present a number to each other which is determined from their value. There is uncertainty in the game as the presented number by a player is determined probabilistically. The upper and lower bounds of the presented number vary depending on the relative relation between the values of the two players. The values of players are determined through evolutionary computation. In our evolutionary computation, each player's value is encoded into an binary string. Various experiments such as different evaluation methods and different experimental settings are performed in a series of computer simulations in order to see if players in our adversary game can successfully evolve.
