Jun Wang
Abstract :
Multi-agent learning arises in a variety of domains where intelligent agents interact not only with the (unknown)
environment but also, critically, with each other. It has an increasing number of applications ranging from
controlling a group of autonomous vehicles to coordinating collaborative bots in production lines, optimising
distributed sensor networks, and machine bidding in competitive e-commerce and financial markets, just to name a
few. Yet, the non-stationary nature calls for a new theory that brings interactions into the learning process. In
this talk, I shall provide an up-to-date introduction on the theory and methods of multi-agent AI, with a focus on
multiagent learning and reasoning framework on the basis of Bayesian decision making. The studies in both game
theory and machine learning will be examined in a unified treatment, making use of variational inference. I shall
also sample our recent work on the subject including mean-field multiagent reinforcement learning, theory of mind
and recursive reasoning, and solution concepts beyond Nash-equilibrium.