AIMS Seminar Series - 13th November 2020
Reinforcement Learning in Multi-Agent Systems
Abstract: Our group at Edinburgh specialises in algorithm development for autonomous decision making in complex environments, with a focus on deep reinforcement learning (RL) applied to multi-agent interaction problems. I will discuss some of the challenges in this area, followed by a selection of ongoing projects in our group which make advances in: multi-agent RL with sparse rewards, ad-hoc teamwork under dynamic team composition, and safe autonomous driving in urban environments. I will conclude by mentioning recent initiatives at the Turing Institute which aim to foster research in multi-agent systems.