Explore the connection between Hamilton-Jacobi-Bellman equations and multi-armed bandit (MAB) problems in this 44-minute seminar talk from the USC Probability and Statistics Seminar series. Delve into the first work establishing this connection in a general setting, as presented by Yuhua Zhu from UCSD. Learn about an efficient algorithm for solving MAB problems based on this newly established link and discover its practical applications. Gain insights into the exploration-exploitation trade-off in sequential decision making under uncertainty, a key concept in MAB paradigms.
Overview
Syllabus
Yuhua Zhu: Continuous-in-time Limit for Bandits (UCSD)
Taught by
USC Probability and Statistics Seminar