Abstract: Reinforcement learning (RL) addresses the design of agents that improve decisions while operating within complex and uncertain environments. In this lecture, I am going to focus on principled and scalable approaches to realizing a range of intelligent learning behaviors. Starting from the fundamentals of RL, the lecture will proceed to cover several important frontiers of contemporary research, such as exploration, state abstraction, and policy optimization, with examples drawn from web services, control, as well as games.
About the Speaker:
Shi Dong is currently a senior researcher at Microsoft, where he is a member of the Cognitive Service Research group. He received his bachelor’s degree from Tsinghua University in 2016, and PhD from Stanford University in 2022. His doctoral research, advised by Prof. Benjamin Van Roy, focuses on using theoretical tools to understand essential elements in reinforcement learning. Besides Microsoft, he has also worked for Google, ByteDance, and DeepMind. One of his works was selected as winner in the 2021 INFORMS George Nicholson Student Paper Competition.
Tencent Meeting（ ID: 935-3935-7603； Passcode:123456 ）
Meeting Link: https://meeting.tencent.com/dm/6xILDn3HGWeV
