Seminar: Transfer Learning, Data Efficiency and Fairness in Deep Reinforcement Learning

Seminar by Matthieu Zimmer, UM-SJTU

Monday, February 8th, 11:00 – 12:00

INRIA Montbonnot Saint-Martin


Abstract: In reinforcement learning, we aim at designing agents that take sequential decisions in unknown environments by learning through their own interaction with such environments. However, learning from scratch is often costly in terms of data to collect, transfer learning can help to overcome some of those difficulties by reusing previously learned knowledge. Jointly, designing more data efficient algorithms is also important to scale to more real world problems. During the presentation, our contributions to these two aspects will be introduced. Besides, as the operations of autonomous systems generally affect simultaneously several users, it is crucial that their designs account for fairness considerations. Hence, we also investigated the problem of learning a policy that treats its users equitably in single and multi-agent cases.

Biography: Matthieu Zimmer received the Ph.D. degree in Computer Science from the University of Lorraine in 2018 within the LORIA Laboratory. Previously, he was with the INRIA, LIP6 and ISIR as research intern, and received the M.S. degree from the University Pierre and Marie Curie. Since 2018, he is a postdoctoral researcher at the Joint Institute of the University of Michigan and Shanghai Jiao Tong University in Shanghai. His researches mainly focus on deep reinforcement learning.