The application of dependency graphs in statistical learning and probabilistic combinatorics

Speaker: Rui Zhang

Abstract

Dependency graphs have been widely used in probability via Stein’s method, cumulants, etc. to establish normal or Poisson approximation, or to obtain concentration results. They are also heavily used in combinatorics, such as the Lovasz local lemma, Janson’s inequality, etc. I will talk about some recent developments regarding dependency graphs, including some concentration bounds and the probability of the non-occurrences of events, by giving applications in statistical learning and random graphs.