IFML Seminar: 09/18/26 - On Robust Quantum Estimation

Abstract: In this talk I will survey recent developments on quantum analogs of robust statistics. Here, the task is to learn quantum states in the presence of potentially adversarial noise: this is arguably even more important than in the classical setting, not only because quantum devices are very noisy, but also for uniquely quantum reasons related to mean-field approximations of many-body systems. As a highlight, I will explain a recent work where we use black-box methods from robust statistics to develop new (semi-)agnostic algorithms for learning product mixed states and product pure states.

Minimal prior quantum background will be expected for this talk---the first half of the talk will be a (hopefully) gentle introduction to the key concepts in quantum learning theory.

Bio: Jerry Li is an associate professor at the University of Washington. His research spans the theoretical foundations of machine learning, high-dimensional statistics, quantum information, and the science of large foundation models. Previously, he was a principal research scientist at Microsoft Research Redmond and a VMware Research Fellow at the Simons Institute for the Theory of Computing. He received his Ph.D. from MIT, where he was advised by Ankur Moitra. His work has been recognized with the 2026 Gödel Prize and MIT’s George M. Sprowls Award for outstanding doctoral research, and has been featured in venues such as the Communications of the ACM and Science.

Zoom Link: https://utexas.zoom.us/j/85330568382