Statistics

Modern machine learning has outstripped attempts to theoretically explain its success from first principles. My research instead takes a modular approach to theory-building: how, and when, can the fundamental building blocks of machine learning be composed to solve more complex problems? My past work has applied this perspective to understand the problem of learning to make reliable sequences of decisions, and to develop algorithms that can prevent the errors in these building blocks from compounding.

Host: Jason Lee

Ph.D. Institution: MIT

Shirshendu Ganguly's research concerns the study of randomness, particularly universality properties in spatially correlated structures. Various phenomena in nature ranging from forest fires, growth of bacteria, behavior of magnetic materials, as well as traffic flow, while superficially different, are expected to exhibit similar fluctuation theories. A particular class of predictions of such behavior was one of the central components of the recent Nobel prize in physics awarded to Giorgio Parisi. However, building a mathematical framework for a rigorous investigation of such models is one of...

My main research interests are in nonparametric and high-dimensional statistics, as well as the statistical foundations of AI. Particular topics include shape-constrained estimation problems; data perturbation methods (e.g. subsampling, bootstrap sampling, random projections, knockoffs); deep learning; in-context learning; nonparametric classification; unconditional and conditional independence testing; estimation of entropy and other functionals; changepoint detection and estimation; missing data; subgroup analysis; variable selection; and applications, including public health, genetics...

My research focuses on probability theory and its connections to statistical physics, studying random structures such as graphs, networks, and geometric surfaces. It investigates processes like random walks, percolation, spanning trees and aims to understand large-scale behavior and phase transitions in complex systems.

Host: Shirshendu Ganguly
Home Institution: Tel Aviv University