This paper analyzes the iterated version of the well-known Boxed Pigs game through Reinforcement Learning. In this scenario, there are two differently sized players (pigs) that compete against each other. The core idea is about sacrificing a pay-off in order to generate some rewards. In our iterated version, these pigs play this game repeatedly using different strategies. We carry out two experiments: in the first one, we train two Q-learning agents against each other to see which equilibrium will be generated. In the second one, we pit the Reinforcement Learning agent against a fixed policy pig. The results of this experiment confirm the ability of Reinforcement Learning techniques in finding the best strategy for maximizing the return independently from the other player choices.      «

This paper analyzes the iterated version of the well-known Boxed Pigs game through Reinforcement Learning. In this scenario, there are two differently sized players (pigs) that compete against each other. The core idea is about sacrificing a pay-off in order to generate some rewards. In our iterated version, these pigs play this game repeatedly using different strategies. We carry out two experiments: in the first one, we train two Q-learning agents against each other to see which equilibrium wi...     »