AI Planning Group, IBM Research
- Date & time
- Monday, February 8, 2021 · 1:00 PM
- Location
- Online (live stream)
Abstract
Influence diagrams (IDs) extend Bayesian networks with decision variables and utility functions to model the interaction between an agent and a system. The standard task is to compute the maximum expected utility (MEU) and optimal policies, one of the most challenging tasks in graphical models. Computing upper bounds on the MEU is desirable because they can guide search or sampling-based methods. In this talk, I present bounding schemes for solving IDs: one extends variational decomposition bounds in marginal MAP, and another is a new submodel tree decomposition method. Empirical results show these bounds are orders of magnitude tighter than previous methods.
About the speaker
Junkyu Lee received his Ph.D. from the CS department at UC Irvine, supervised by Rina Dechter, and is currently a resident at the IBM Research AI planning group. His research focuses on graphical model inference and heuristic search for sequential decision making under uncertainty.