Reinforcement Learning Produces Dominant Strategies for the Iterated Prisoner's Dilemma