Structured Learning Team Seminar (Talk by Lingwei Zhu, University of Alberta).

This is an online seminar. Registration is required.
【Structured Learning Team】
【Date】2024/April/16(Tue) 9:00-10:00(JST)
*【Speaker】Dr. Lingwei Zhu, University of Alberta *

Title: Entropy Regularization for Practical Real-World Applications of Reinforcement Learning

Abstract:
The reinforcement learning community has seen an unprecedented increase in popularity in the past decade. However, despite that numerous methods have been developed and proposed, its real-world applications remain very limited, with most of them restricted to simulation. In this talk, I will identify practical concerns of reinforcement learning when applied to real-world complicated systems, and talk about my solutions to tackle these challenges. Specifically, entropy regularization plays a central role in these methods on how to learn efficiently, safely and robustly. The included applications range from industrial manufacturing plants to simulation-to-reality robotics.