Defining and Mitigating Collusion in Multi-Agent Systems
Keywords: collusion, multi-agent systems, multi-agent reinforcement learning, mechanism design
TL;DR: We define collusion in partially observable stochastic games and investigate three kinds of intervention to mitigate it.
Abstract: Collusion between learning agents is increasingly becoming a topic of concern with the advent of more powerful, complex multi-agent systems. In contrast to existing work in narrow settings, we present a general formalisation of collusion between learning agents in partially-observable stochastic games. We discuss methods for intervening on a game to mitigate collusion and provide theoretical as well as empirical results demonstrating the effectiveness of three such interventions.
Submission Number: 7
Loading