Hypothesis Testing in High-Dimensional Regression under the Gaussian Random Design Model

Andrea Montanari
Stanford University

A large amount of work has been devoted, over the last ten years, to high-dimensional statistical estimation problems. Examples range from signal processing to collaborative filtering. The methods developed are mostly based on convex optimization techniques, and typically do not provide tools to reason about uncertainty or confidence. Hypothesis testing provides a classical framework for addressing this problem. I will discuss the challenges related to hypothesis testing in this context. [Joint work with Adel Javanmard]

Back to Structure and Randomness in System Identification and Learning