Event Date
Event Date
Location
Mathematical Sciences Building 1147
Speaker: Jasper Lee, Assistant Professor, Department of Computer Science, UC Davis
Title: "Mean estimation in 1 and high dimensions"
Abstract: Even for statistical problems as fundamental as mean estimation, there is a theory-practice divide. Conventional methods like the sample mean, while supported by theoretical results under strong assumptions, are often brittle in the presence of extreme data. Practitioners thus often use ad-hoc and unprincipled "outlier removal" heuristics, revealing a marked theory-practice gap even for this very basic problem.
In this talk, I will describe my work bridging such theory-practice divide. I will specifically highlight 3 works: A) constructing a statistically-optimal and computationally-efficient 1-dimensional mean estimator, whose estimation error is optimal even in the leading multiplicative constant, under bare minimum distributional assumptions, B) a rather different but optimal mean estimator for the "very high-dimensional" regime, and C) a recent result showing that the estimator from "A)" is robust even under the presence of adversarial data corruption.
Bio: Jasper Lee is an assistant professor at the University of California, Davis, in the Department of Computer Science and the Graduate Group in Applied Mathematics. He completed his PhD at Brown University, advised by Paul Valiant, and was subsequently a postdoc mentored by Ilias Diakonikolas at UW Madison.
His research interests are broadly in the foundations of data science, aiming to design practical, data-efficient and computationally-efficient algorithms for a variety of statistical applications.
Faculty webpage (UC Davis): https://cs.ucdavis.edu/people/jasper-lee