Emergent Discrete Communication in Semantic Spaces
This work addresses the challenge of improving multi-agent communication for AI systems, offering a novel approach that enhances zero-shot understanding and robustness, though it is incremental in building on existing embedding techniques.
The paper tackles the problem of enabling neural agents to communicate via discrete tokens that are semantically meaningful, showing that their method allows agents to cluster tokens in meaningful ways and outperform one-hot communication in noisy environments and in human-agent interaction.
Neural agents trained in reinforcement learning settings can learn to communicate among themselves via discrete tokens, accomplishing as a team what agents would be unable to do alone. However, the current standard of using one-hot vectors as discrete communication tokens prevents agents from acquiring more desirable aspects of communication such as zero-shot understanding. Inspired by word embedding techniques from natural language processing, we propose neural agent architectures that enables them to communicate via discrete tokens derived from a learned, continuous space. We show in a decision theoretic framework that our technique optimizes communication over a wide range of scenarios, whereas one-hot tokens are only optimal under restrictive assumptions. In self-play experiments, we validate that our trained agents learn to cluster tokens in semantically-meaningful ways, allowing them communicate in noisy environments where other techniques fail. Lastly, we demonstrate both that agents using our method can effectively respond to novel human communication and that humans can understand unlabeled emergent agent communication, outperforming the use of one-hot communication.