Bayesian Action Decoder for Deep Multi-Agent Reinforcement Learning
This addresses the problem of enabling effective strategy discovery and communication in multi-agent systems for AI researchers, though it is incremental as it builds on existing multi-agent reinforcement learning frameworks.
The paper tackles the challenge of scalable multi-agent reinforcement learning in complex, partially observable settings by introducing the Bayesian action decoder (BAD), which uses an approximate Bayesian update to form a public belief from agents' actions. It outperforms policy gradient methods in a matrix game and achieves state-of-the-art results in the cooperative card game Hanabi, surpassing all prior learning and hand-coded approaches.
When observing the actions of others, humans make inferences about why they acted as they did, and what this implies about the world; humans also use the fact that their actions will be interpreted in this manner, allowing them to act informatively and thereby communicate efficiently with others. Although learning algorithms have recently achieved superhuman performance in a number of two-player, zero-sum games, scalable multi-agent reinforcement learning algorithms that can discover effective strategies and conventions in complex, partially observable settings have proven elusive. We present the Bayesian action decoder (BAD), a new multi-agent learning method that uses an approximate Bayesian update to obtain a public belief that conditions on the actions taken by all agents in the environment. BAD introduces a new Markov decision process, the public belief MDP, in which the action space consists of all deterministic partial policies, and exploits the fact that an agent acting only on this public belief state can still learn to use its private information if the action space is augmented to be over all partial policies mapping private information into environment actions. The Bayesian update is closely related to the theory of mind reasoning that humans carry out when observing others' actions. We first validate BAD on a proof-of-principle two-step matrix game, where it outperforms policy gradient methods; we then evaluate BAD on the challenging, cooperative partial-information card game Hanabi, where, in the two-player setting, it surpasses all previously published learning and hand-coded approaches, establishing a new state of the art.