Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 55 - Advances in Bayesian Sparse Regression
Type: Contributed
Date/Time: Monday, August 3, 2020 : 10:00 AM to 2:00 PM
Sponsor: Section on Bayesian Statistical Science
Abstract #313737
Title: A Generalized Likelihood Based Bayesian Approach for Scalable Joint Regression and Covariance Estimation in High Dimensions
Author(s): Srijata Samanta* and Kshitij Khare and George Michailidis
Companies: University of Florida and University of Florida and University of Florida
Keywords: high-dimensional data; Bayesian variable selection; generalized likelihood; multiple response linear regression
Abstract:

We consider a high-dimensional multiple response linear regression model with the focus being joint estimation of the regression and the dependence structure among the variables. We propose a generalized likelihood based Bayesian approach for computationally efficient joint estimation of the sparsity patterns in regression coefficients matrix B and the inverse covariance ? of the response variables. A procedure for obtaining parameter estimates of B and ? consistent with these sparsity patterns is also proposed. Finally we establish high-dimensional consistency of the proposed procedure for estimation of sparsity patterns of B and ? when the number of response variables and the number of predictors grow nearly exponentially with the sample size and also examine the performance of the approach based on synthetic data.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2020 program