Advances in Bayesian variable selection
Abstract Number:
1510
Submission Type:
Topic-Contributed Paper Session
Participants:
Brian Reich (1), Brian Reich (1), Howard Bondell (2), Nadja Klein (3), Leo Duan (4), Arkaprava Roy (4), Jonathan Williams (1)
Institutions:
(1) North Carolina State University, N/A, (2) University of Melbourne, N/A, (3) Humboldt-Universitaet zu Berlin, N/A, (4) University of Florida, N/A
Chair:
Session Organizer:
Speaker(s):
Session Description:
Variable selection is a critical step in any regression analysis and many Bayesian methods are now available for this fundamental task. However, many unsolved issues remain including computational methods to deal with massive datasets, extending variable selection to non-linear and non-Gaussian models and establishing theoretical properties of existing and new approaches. This session includes five speakers that range from junior faculty to established leaders in the field, and includes two international speakers. Howard Bondell will introduce new methods that avoid specifying a likelihood function and therefore are more robust to model misspecficiation than standard approaches. Nadja Klein and Arkaprova Roy will extend linear regression methods to more difficult settings including non-linear and dependent data. Leo Duan will show how of perform Bayesian variable selection with a massive number of predictors and Jonathan Williams will study the theoretical properties of Bayesian methods. Therefore, this session represents a comprehensive view of recent developments in the area of Bayesian variable selection. List speakers:
(1) Howard Bondell (University of Melbourne): Shrinkage and selection for high-dimensional Bayesian estimating equations
(2) Nadja Klein (Technical University Dortmund): Spike-and-slab group horseshoe for Bayesian effect selection in additive regression models
(3) Leo Duan (University of Florida): Taming combinatorial explosion with new optimization-induced priors
(4) Arkaprava Roy (University of Florida): Bayesian semiparametric dynamic tensor factor models for high-dimensional multi-group longitudinal neuroimaging data
(5) Jonathan Williams (North Carolina State University): Epsilon admissible subset approaches to Bayesian variable selection problems
Sponsors:
International Society for Bayesian Analysis (ISBA) 1
International Statistical Institute 3
Section on Bayesian Statistical Science 2
Theme:
Statistics and Data Science: Informing Policy and Countering Misinformation
No
Applied
No
Estimated Audience Size
Medium (80-150)
I have read and understand that JSM participants must abide by the Participant Guidelines.
Yes
I understand and have communicated to my proposed speakers that JSM participants must register and pay the appropriate registration fee by June 1, 2024. The registration fee is nonrefundable.
I understand
You have unsaved changes.