This talk is part of the NLP Seminar Series.

Multi-Agent Reinforcement Learning towards Zero-Shot Emergent Communication

Kalesha Bullard, Facebook AI Research
Date: 10:00am - 11:00am PT, May 27 2021
Venue: Zoom (link hidden)


Effective communication is an important skill for enabling information exchange and cooperation in multi-agent settings, in which AI agents coexist in shared environments with other agents (artificial or human). Indeed, emergent communication is now a vibrant field of research, with common settings involving discrete cheap-talk channels. One limitation of this setting however is that it does not allow for the emergent protocols to generalize beyond the training partners. Furthermore, the typical problem setting of discrete cheap-talk channels may be less appropriate for embodied agents that communicate implicitly through physical action. This talk presents research that investigates methods for enabling AI agents to learn general communication skills through interaction with other artificial agents. In particular, the talk will focus on my ongoing work within Multi-Agent Reinforcement Learning, investigating emergent communication protocols, inspired by communication in more realistic settings. We present a novel problem setting and a general approach that allows for zero-shot communication (ZSC), i.e., emergence of communication protocols that can generalize to independently trained agents. We also explore and analyze specific difficulties associated with finding globally optimal ZSC protocols, as complexity of the communication task increases or the modality for communication changes (e.g. from symbolic communication to implicit communication through physical movement, by an embodied artificial agent). Overall, this work opens up exciting avenues for learning general communication protocols in more complex domains.


Kalesha Bullard is a postdoctoral researcher at Facebook AI Research. She completed her PhD in Computer Science at Georgia Institute of Technology in 2019, where her research focused within the space of interactive robot learning. During her postdoc, Kalesha has expanded her research to explore the space of multi-agent reinforcement learning, currently investigating how to enable embodied multi-agent populations to learn general communication protocols. More broadly, Kalesha’s research interests span autonomous reasoning and decision making for artificial agents in multi-agent settings. To date, her research has focused on principled methods for enabling agents to learn through interaction with other agents (human or artificial) to achieve shared goals. Beyond research, Kalesha has participated in a number of service roles throughout her research career, recently serving on organizing and program committees for workshops associated with several top Artificial Intelligence conference venues (e.g. NeurIPS, AAAI, AAMAS). This past year, she was selected as one of the 2020 Electrical Engineering and Computer Science (EECS) Rising Stars, hosted by UC Berkeley.