Department Seminar Series: Bodhisattva Sen, Associate Professor, Department of Statistics, Columbia University

Abstract: I will introduce and motivate nonparametric estimation of a multivariate convex regression function and describe some recent advances in this topic. Least squares estimation of the multivariate convex regression function will be discussed along with an application to testing goodness-of-fit against linear models. Adaptation properties of the convex least squares estimator will be illustrated via finite sample risk bounds. The talk will end with some ongoing and
open problems in this area.