David Dunson named among top scientists in the field of mathematics

David Dunson banner image showing David is named top mathematician

David Dunson, Arts & Sciences Professor of Statistical Science, has been named among top scientists in the field of mathematics for the 2023 ranking of researchers at Research.com

David Dunson’s scholarly influence combines theoretical and applied research in analyzing complex and high-dimensional data while characterizing uncertainty in inferences through probability distributions via Bayesian methods. He has made pioneering research contributions with a rich variety of popular statistical inference methods based on flexible nonparametric and robust models. These methods are often developed in challenging applications such as environmental health, biodiversity research, neuroscience, and epidemiology.  

In environmental health, Dunson focuses on quantifying health effects of multiple different chemical exposures, like air pollution, which may interact and vary in their effects over time. In biodiversity research, his focus is on understanding how complex interactions between communities and networks of species are shaped by environmental conditions and climate change.  In neuroscience, his emphasis has been on developing novel inferential tools for analyzing human brain connection networks and their relationships with human traits. 

In these applications, Dunson develops probabilistic inferential models that are designed to handle a broad variety of data types, ranging from networks, to functions, to immensely long binary and count vectors indicative of species sampling and omics data.  

Given the tendency for data to be both very high-dimensional and sparse in these areas of applied research, a key focus of Dunson’s theoretical research has been on designing and investigating latent structure models that characterize complex interrelationships in data while enabling dimensionality reduction. A second broad emphasis of his work is developing scalable computational tools that allow drawing inference with these complex probability models from extremely large data sets encountered in practical applications.