Presenter:  Professor Alexander Volfovsky, Duke Univerisity 

Description: Bayesian Additive Regression Trees have become ubiquitous across Bayesian statistics in the last decade. Leveraging a sum-of-trees approach, BART is able to flexibly model or approximate E[Y|x], without too much concern for overfitting. Recent advances in BART have seen its extension to binary outcomes, variance regression and to survival analysis. In this work we provide a natural extension of the BART framework to incorporating latent variables and demonstrate how this allows us to study density regression problems and to capture potentially hierarchical structure in our data. For our Density RegressionBART (DR-BART) we prove that the posterior induced by our model concentrates quickly around true generative functions that are sufficiently smooth. We also analyze DR-BART's performance on a set of challenging simulated examples, where it outperforms various other methods for Bayesian density regression. We conclude by demonstrating our generalized Latent Variable BART (LV-B! ART) framework in the context of hierarchical modeling and sensitivity analysis to unobserved confounding in observational studies. 

Bio: Alexander Volfovsky is an Associate Professor of Statistical Science at Duke University. He works primarily on fundamental properties in network analysis, causal inference and computational social science. He is the Co-Director two labs at Duke: (1) The Polarization Lab at Duke which brings together scholars from the social sciences, statistics, and computer science to develop new technology to bridge America’s partisan divide. (2) The Almost Matching Exactly Lab at Duke which brings together statisticians, computer scientists, economists and political scientists to develop tools for interpretable causal inference. His work has appeared in various academic journals including Journal of the American Statistical Association, the Journal of Machine Learning Research, the Annals of Statistics, and the Proceedings of the National Academy of Sciences, and has been recognized with the Paul Lazarsfeld Award from the APSA and best paper awards from sections of the ASA. His current ! research is supported by awards from the NSF, the Templeton Foundation, Facebook and Google. 

Hosted by:  Professor Sheng Jiang

Zoom link:  https://ucsc.zoom.us/j/94492730062?pwd=ZTkyZGRiS0NUelhxWVd6L2NVanAwUT09