Sam Hopkins (UC Berkeley)Sep 11, 2020 Title and AbstractRecent Advances in Algorithmic Heavy-Tailed Statistics However, most of the high-dimensional estimators known to obtain such sharp confidence intervals under weak assumptions appear computationally intractable — requiring running time exponential in dimension. I will discuss several recent works giving polynomial-time algorithms with similar guarantees, focusing on new algorithms for high-confidence covariance estimation and linear regression in joint work with Cherapanamjeri, Kathuria, Raghavendra, and Tripuraneni, STOC 2020. BioSam Hopkins is a Miller Postdoctoral Fellow at UC Berkeley in the department of Electrical Engineering and Computer Science. He works on algorithms and computational complexity, especially for computationally challenging problems in high-dimensional statistics. He obtained a PhD from Cornell University, where his work was supported by a Microsoft Graduate Fellowship and an NSF Graduate Research Fellowship, and a BS from the University of Washington. In 2021 he will join MIT as an assistant professor of computer science |