SDS Seminar Series – Abhra Sarkar, University of Texas at Austin

The Fall 2024 SDS Seminar Series continues on September 27th from 2:00 p.m. to 3:00 p.m. with Dr. Abhra Sarkar (Associate Professor, Department of Statistics and Data Sciences, University of Texas at Austin). This event is in-person in the Avaya Room (POB 2.302).    

Title: (Bayesian) Semiparametric Local Inference (and Other Stories)

Abstract: In this talk, I will present some ideas on flexible (Bayesian) semiparametric local inference models in various settings. Here I define the problem of ‘local inference’ broadly as one where different sets of predictors can influence a response variable differently in different parts of a domain of interest (often continuous but not always). For instance, in longitudinal experiments, predictors may have varying effects on the response over time, while in spatial settings, their influences may differ across locations. Although the basic notion is not new to statistics, developing rigorous modeling, computational, and theoretical frameworks for local inference in complex real-world settings, particularly for categorical predictors such as experimental or demographic factors in laboratory and survey settings, poses significant challenges, and the literature remains sparse. I have encountered interesting local inference problems essentially in all application areas I have worked on, including behavioral neuroscience, imaging neuroscience, nutritional epidemiology, and so forth. Often domain-specific knowledge introduces additional unique nuances and twists, making them (at least from a personal perspective) even more fascinating. In the interest of time, I will focus on the first two application areas mentioned above. Along the way, I’ll also share some stories about how these ideas have developed during my time here at SDS.