본문 바로가기

Variational Inference Aided Variable Selection For Spatially Structured High Dimensional Covariates

문서 유형
연구출판물 · 학술지 논문
등재
SCIEScopus
저널 순위
JCR Q1Scopus Q1
피인용
Web of Science 0Scopus 0
FWCI (Scopus)
0
JCR IF
JCR 2024 1.8
ISSN
1061-8600

저자

  1. Nandy, Siddhartha제1저자Case Western Reserve Univ, Dept Math Appl Math & Stat, Cleveland, OH USA
  2. 김민우Pusan Natl Univ, Dept Stat, Busan, South Korea
  3. Bhattacharya, Shrijita교신저자Michigan State Univ, Dept Stat & Probabil, 619 Red Cedar Rd Wells,Room C419, E Lansing, MI 48824 USA
  4. Maiti, TapabrataMichigan State Univ, Dept Stat & Probabil, 619 Red Cedar Rd Wells,Room C419, E Lansing, MI 48824 USA

초록

We consider the problem of Bayesian high dimensional variable selection in linear regression when a spatial structure exists among the covariates. We use an Ising prior to model the structural connectivity of the covariates with an undirected graph and the connectivity strength with Ising distribution parameters. Ising models, which originated in statistical physics, are widely used in computer vision and spatial data modeling. Although a Gibbs solution to this problem exists, the solution involves the computation of determinants and inverses of high dimensional matrices, rendering it unscalable to higher dimensions. Furthermore, the lack of theoretical support limits this important tool's use for the broader community. This article proposes a variational inference-aided Gibbs approach that enjoys the same variable recovery power as the standard Gibbs solution while being computationally scalable to higher dimensions. We establish strong selection consistency for our proposed approach, along with its competitive numerical performance under varying simulation scenarios. Supplementary materials for this article are available online.

키워드

Bayesian variable selectionHigh-dimensionalIsing distributionStructured covariatesVariational bayesISING-MODELREGRESSION

식별자

Scopus EID
2-s2.0-105019109616
WoS UID
001592445000001