STATISTICS COLLOQUIUM
Snigdhansu Chatterjee
Director, Institute for Research in Statistics and its Applications (IRSA)
Professor, School of Statistics
University of Minnesota
Bayesian equation selection and statistical learning of stochastic dynamical systems
Abstract
Natural processes adhere to laws of science, that are often describable as systems of (partial, stochastic) differential equations, or stochastic dynamical systems. We present a Bayesian framework for discovering such dynamical systems under assumptions that align with real-life scenarios, including the availability of relatively sparse data. We investigate different modeling and computational strategies that may be used for teasing out the important details about the dynamical system. These include direct Bayesian modeling of the observed data, functional data approaches and Bayesian deep learning approaches. The proposed framework can be used for evaluation and validation of scientific and computational models of complex natural phenomena, and in turn, for obtaining precise and accurate predictions in tasks related to precision medicine, climate modeling and other applications. We present theoretical and methodological details and examples from two use cases: on modeling cancer and on modeling climate.
Bio: Dr. Chatterjee is Professor of Statistics and the Director of the Institute for Research in Statistics and its Applications (IRSA) at the University of Minnesota. He is also a fellow of the Institute on the Environment and member of the Minnesota Population Center at the University of Minnesota. His research interests include statistical foundations of data science, geometry of high dimensional data, Bayesian statistics, applications of statistics, machine learning to several domains. Dr. Chatterjee graduated from the Indian Statistical Institute in 2000.
Join by meeting number |
Meeting number (access code): 2624 609 2570 |
Meeting password: iHpcU6Tcx27 |
For more information, contact: Tracy Burke at tracy.burke@uconn.edu