Speaker |
Zoltan Szabo |
|---|---|
Affiliation |
London School of Economics and Political Science |
Date |
Wednesday, 05 November 2025 |
Time |
12:30-13:30 |
Location |
Ground floor lecture theatre, Sainsbury Wellcome Center, 25 Howland St, W1T 4JG |
Link |
https://ucl.zoom.us/j/99748820264 |
Event series |
Jump Trading/ELLIS CSML Seminar Series |
Abstract |
Kernel Stein discrepancies are among the most powerful approaches to quantify goodness-of-fit on a wide variety of domains with numerous successful applications. To our best knowledge, all available KSD estimators achieve root-n-convergence. We present (using two different proof techniques) matching lower bounds both on R^d and on general domains, providing complementary insights. This is joint work with Jose Cribeiro-Ramallo, Agnideep Aich, Florian Kalinke, and Ashit Baran Aich; preprint: https://arxiv.org/abs/2510.15058. |
Biography |
Zoltán Szabó is a Professor of Data Science in the Department of Statistics at the London School of Economics and Political Science (LSE). His research focuses on statistical machine learning, particularly kernel methods, information-theoretic estimators (ITE), and scalable computation. He applies these methods to a wide range of domains, including safety-critical learning, style transfer, shape-constrained prediction, hypothesis testing, distribution regression, dictionary learning, structured sparsity, independent subspace analysis, Bayesian inference, and various applied areas such as finance, economics, climate data analysis, criminal data analysis, collaborative filtering, emotion recognition, face tracking, remote sensing, natural language processing, and genomics. |