Multi-agent reinforcement learning is a logical extension of a single agent task. However, even though we had great successes in single agent deep reinforcement learning in recent years, a multi-agent extension does encounter some problems currently not dealt with.

At the seminar, we will review the current state of the field of multi-agent reinforcement learning. We will discuss various multi-agent tasks, see, how deep learning methods and multi-agent reinforcement learning complement each other. We will also go through state-of-the-art algorithms and some unsolved problems.

Speaker: Anna Fedorova.

Presentation language: Russian.

Date and Time: January 22nd, 18:30-20:00.

Place: Times, room 204.