Title: Spectral Methods for signal recovery in multi-modal problems

Speaker: Subhabrata Sen, Associate Professor in the Department of Statistics at Harvard University.

Time: Wednesday, October 28, 3:30pm,  Austin 313.

Link: https://teams.microsoft.com/meet/278035717198411?p=7rpRj5tHFgKYCtcTVV

Abstract: Motivated by multi-modal estimation, we will study a multi-view spiked random matrix model in which several noisy matrix observations contain correlated spikes. We will introduce a new spectral method for signal recovery in this setting and characterize the behavior of the top eigenvalue and the associated eigenvector. Our results will identify the sharp threshold for signal recovery in this problem. We will discuss broader implications for multi-modal data analysis. Based on joint work with Xiaodong Yang and Yue M. Lu.

Bio of Subhabrata Sen: Subhabrata Sen is an Associate Professor in the Department of Statistics at Harvard University. His research focuses on the foundations of high-dimensional learning and inference. He is especially interested in the fundamental limits of inference from structured data, including networks, multi-view and multi-modal data, and models with dependence. A central goal of his research is to develop new algorithms that are practically impactful and explain the empirical success of modern machine learning algorithms.