Sequential strategy for learning multi-stage multi-agent collaborative games

Abstract

An alternative approach to learning decision strategies in multi-state multiple agent systems is presented here. The method, which uses a game theoretic construction of “best response with error” does not rely on direct communication between the agents in the system. Limited experiments show that the method can find Nash equilibrium points at least for a 2 player multi-stage coordination game and converges more quickly than a comparable co-evolution method.