Date: Wednesday, May 5, 2021
Time: 12:00 – 1:00 pm Pacific Time
Location (Zoom): https://berkeley.zoom.us/j/91562514796

Abstract:
The past few years have seen the growth of deep reinforcement learning (RL) as a new and powerful optimization technique. This thesis looks at deep RL from the systems perspective in two ways: how to design systems that scale the computationally demanding algorithms used by researchers and practitioners, and conversely, how to apply deep RL to expand the state of the art in systems. The first half of this talk overviews the design and evolution of RLlib, a scalable and widely adopted open source library for distributed reinforcement learning. RLlib offers new abstractions RL practitioners can leverage to build their own distributed RL algorithms, as well as an architecture that enables support for RL workloads in a flexible and high-performance way. The second half of this talk looks at how we can move towards leveraging RL and ML for improving systems, specifically diving into examples improving the construction of packet classification trees and database cardinality estimators.