Geometry Seminar

Date: November 8, 2021

Time: 3:00PM - 4:00PM

Location: zoom

Speaker: V. Makam

Title: Emerging applications of invariant theory to statistics

Abstract: Maximum likelihood estimation is a technique in statistics that is widely used to recover the probability distribution in a statistical model that best explains the empirical data. A curious connection between stability notions in invariant theory and maximum likelihood estimation for a large class of statistical models was uncovered by Amendola, Kohn, Reichenbach, and Seigal, with recent results in complexity theory forming a bridge. In this talk, I will give an overview of these connections, present some exciting results in a few different settings and sketch out the potential future directions. I will be presenting joint work (and work in progress) combining a few projects with Gergely Berczi, Harm Derksen, Cole Franks, Eloise Hamilton, Philipp Reichenbach, Anna Seigal, and Michael Walter.