Computer Science Department
Iterated Prisoner’s Dilemma Tournament
Prisoner’s Dilemma is a two-player game in which each player has two options. The benefit of each option to one player is contingent on the decision made by the opposing player. The dilemma arises because both players reveal their decisions simultaneously, so players can never be certain of their payoff until after they commit to a decision. This game has been studied extensively in a number of fields, including computer science, psychology, economics, and politics, due to the widespread applicability of the basic premise of the game to real-world problems.
This project focuses on examining Prisoner’s Dilemma from the standpoints of learning and artificial intelligence. Due to the simplicity of the game, it is very easy to pick up and start playing. But it also allows for some very sophisticated strategies to form. First, a Prisoner’s Dilemma tournament platform will be created. Then, artificial intelligence players of the game will be created. Some will have set strategies taken from the literature, but one AI player will be created with an unsupervised learning algorithm. It will play against the other preset strategies and will create its own strategy based on those experiences.