Universally Expressive Communication in Multi-Agent Reinforcement Learning
This addresses a foundational limitation in multi-agent systems for AI, though it is incremental by building on graph neural network theory.
The paper tackles the problem of whether communication protocols in multi-agent reinforcement learning can express arbitrary policies, showing that augmenting agent observations with unique IDs and random noise yields universally expressive communication and improves performance on tasks requiring such expressiveness.
Allowing agents to share information through communication is crucial for solving complex tasks in multi-agent reinforcement learning. In this work, we consider the question of whether a given communication protocol can express an arbitrary policy. By observing that many existing protocols can be viewed as instances of graph neural networks (GNNs), we demonstrate the equivalence of joint action selection to node labelling. With standard GNN approaches provably limited in their expressive capacity, we draw from existing GNN literature and consider augmenting agent observations with: (1) unique agent IDs and (2) random noise. We provide a theoretical analysis as to how these approaches yield universally expressive communication, and also prove them capable of targeting arbitrary sets of actions for identical agents. Empirically, these augmentations are found to improve performance on tasks where expressive communication is required, whilst, in general, the optimal communication protocol is found to be task-dependent.