Using Multi-Arm Bandit Algorithm for Changing Background Color in Computer Games

Game designers are always looking for ways to identify their audience s preferences and favorites. In fact, they want to design those games that be more attractive and players can interact with them. In any game, there are some variables that can be initialized to attract audiences and improve games. In this work, we pay to background color as a variable of games and using Multi-Arm Bandit (MAB) approach to determine this variable. Our purpose is to demonstrate how MAB can be used to change automatically the background color of the game, according to player function in the game. For implementing Multi-Arm bandit, we use epsilon-greedy algorithm. To survey the correctness and effectiveness of this method, we use a user study and test a particular population of players. The results show that MAB can detect the player s favorite color (based on commenting player before play game and with a questionnaire) in most cases. Our defined metric for algorithm to detect and choose the best color is the score of playing the game. The statistics results specify that this method can be applicable and useful to improve the games and determine all variables of them

کلیدواژه ها:

Multi-arm Bandit ، Background color of Game ، Game Design