Abbas Mehrabian - McGill University

Title: Learning mixtures of Gaussians

Abstract:

Distribution learning lies at the intersection of statistics, theoretical computer science and machine learning. We give an overview of this area, with the problem of learning mixtures of high-dimensional Gaussians as a running example. In particular, we prove nearly tight sample complexity bounds for this problem in the density estimation model.

Based on joint work with Hassan Ashtiani, Shai Ben-David, Luc Devroye, Nick Harvey, Christopher Liaw, Yani Plan, and Tommy Reddad.