Colloquium Series: Ashwin Pananjady

Our upcoming event for the Statistics Colloquium Series is scheduled for Monday, November 9th  from 12:00 – 1:00pm (ET) and will be an in-person presentation at Maxwell-Dworkin, Room 119. Lunch will be provided to guests following the talk. This week's speaker will be Ashwin Pananjady, Gerald D. McInvale Early Career Assistant Professor at Georgia Tech.

Variational inequality approaches to imaging and missing data problems

This talk will present two new (stochastic) variational inequality (VI) problems inspired by applications in statistical signal processing and data science.  The first is a new VI-based method for single material recovery in polychromatic computed tomography. The method is the first provable recovery algorithm in this problem, bypassing prior approaches based on nonconvex optimization. It is also fast, accurate and flexible in practice, and can produce reconstructions that are comparable in quality to state-of-the-art algorithms while reducing radiation doses by two orders of magnitude. The second problem is classification with missing data. When the covariate distribution is unknown, most approaches to this classical problem (even in the linear setting and with covariates missing completely at random) are based on nonconvex optimization, and largely heuristic. I will present a VI-based method that achieves provable signal recovery at parametric rates, and even in high-dimensional settings. I will also present some new information-theoretic lower bounds that depend on the missingness mechanism and reveal several nonstandard phenomena in this problem.

Speaker's Bio:

Ashwin Pananjady is the Gerald D. McInvale Early Career Assistant Professor at Georgia Tech, with a joint appointment between the H. Milton Stewart School of Industrial and Systems Engineering and the School of Electrical and Computer Engineering. His research interests lie broadly in statistics, optimization, signal processing and information theory, as well as their applications in data science, machine learning, and reinforcement learning. Pananjady received his Ph.D. from the Department of Electrical Engineering and Computer Sciences (EECS) at the University of California Berkeley and the B.Tech. degree (with honors) in Electrical Engineering from the Indian Institute of Technology Madras. His accolades include the David J. Sakrison Memorial Prize from UC Berkeley for the best dissertation in EECS, a Simons-Berkeley Research Fellowship, early career recognitions from the Bernoulli Society, Institute of Mathematical Statistics and Google Research, paper prizes from the Mathematical Optimization Society and Algorithmic Learning Theory conference, and the R. Narasimhan Memorial Lecture Award from TIFR.