Activity Number:
|
312
- Bayesian Variable Selection: When Horseshoe Meets Nonlocal
|
Type:
|
Topic Contributed
|
Date/Time:
|
Wednesday, August 5, 2020 : 10:00 AM to 11:50 AM
|
Sponsor:
|
Section on Bayesian Statistical Science
|
Abstract #312817
|
|
Title:
|
Nonlocal Functional Priors for Nonparametric Hypothesis Testing and High-Dimensional Model Selection
|
Author(s):
|
Minsuk Shin* and Anirban Battacharya
|
Companies:
|
University of South Carolina and Texas A&M University
|
Keywords:
|
Bayesian variable selection;
High-dimensional inference;
Nonparametric regression
|
Abstract:
|
In this talk, we introduce a new class of prior densities that can be used for nonparametric Bayesian hypothesis testing and model selection. Members of this class of prior densities are called non-local functional prior (NLfP) densities. We demonstrate that NLfP densities provide improved convergence rates in many nonparametric Bayesian hypothesis tests. When used to perform high-dimensional Bayesian model selection for nonparametric additive models, we show that the resulting procedures are consistent, and find that the model selection procedures based on NLfP densities outperform existing state-of-the-art alternatives in several standard simulation scenarios and real data sets.
|
Authors who are presenting talks have a * after their name.