The Statistics Seminar speaker for March 25, 2015 will be Karthik Sridharan from Cornell's Computer Science department.
Title: Learning and Estimation with Square Loss: Offset Rademacher Complexity and the Star Algorithm
Abstract: We consider the problem of regression with square loss and general classes of regression functions. We introduce a notion of offset Rademacher complexity that provides a transparent way to study localization both in expectation and in high probability and provide estimation and learning rates for the problem. For any (possibly non-convex) class, the excess loss of a two-step estimator is shown to be upper bounded by this offset complexity through a novel geometric inequality. In the convex case, the estimator reduces to an empirical risk minimizer. The rates recovered shows interesting connections between statistical estimation in the well specified case and statistical learning with square loss (estimation with model misspecification).
Refreshments will be served after the seminar in 1181 Comstock Hall.