Using Multi-Agent Reinforcement Learning in Auction Simulations

Abstract: Game theory has been developed by scientists as a theory of strategic interaction among players who are supposed to be perfectly rational. These strategic interactions might have been presented in an auction, a business negotiation, a chess game, or even in a political conflict aroused between different agents. In this study, the strategic (rational) agents created by reinforcement learning algorithms are supposed to be bidder agents in various types of auction mechanisms such as British Auction, Sealed Bid Auction, and Vickrey Auction designs. Next, the equilibrium points determined by the agents are compared with the outcomes of the Nash equilibrium points for these environments. The bidding strategy of the agents is analyzed in terms of individual rationality, truthfulness (strategy-proof), and computational efficiency. The results show that using a multi-agent reinforcement learning strategy improves the outcomes of the auction simulations.