Quasi-Equivalence Discovery for Zero-Shot Emergent Communication

TL;DR

This paper introduces the Quasi-Equivalence Discovery (QED) algorithm, enabling zero-shot coordination in emergent communication by discovering symmetries and optimizing protocols that generalize across independently trained agents, especially in costly communication scenarios.

Contribution

The paper proposes QED, a novel algorithm that automatically discovers symmetries to achieve zero-shot emergent communication without prior symmetry knowledge.

Findings

QED effectively discovers symmetries in referential games.

QED achieves optimal zero-shot coordination policies.

QED outperforms methods requiring prior symmetry access.

Abstract

Effective communication is an important skill for enabling information exchange in multi-agent settings and emergent communication is now a vibrant field of research, with common settings involving discrete cheap-talk channels. Since, by definition, these settings involve arbitrary encoding of information, typically they do not allow for the learned protocols to generalize beyond training partners. In contrast, in this work, we present a novel problem setting and the Quasi-Equivalence Discovery (QED) algorithm that allows for zero-shot coordination (ZSC), i.e., discovering protocols that can generalize to independently trained agents. Real world problem settings often contain costly communication channels, e.g., robots have to physically move their limbs, and a non-uniform distribution over intents. We show that these two factors lead to unique optimal ZSC policies in referential games, where agents use the energy cost of the messages to communicate intent. Other-Play was recently introduced for learning optimal ZSC policies, but requires prior access to the symmetries of the problem. Instead, QED can iteratively discovers the symmetries in this setting and converges to the optimal ZSC policy.