Large-scale Bayesian Structure Learning for Gaussian Graphical Models using Marginal Pseudo-likelihood
Authors:
Preprint
Abstract
Bayesian methods for learning Gaussian graphical models offer a robust framework that addresses model uncertainty and incorporates prior knowledge. Despite their theoretical strengths, the applicability of Bayesian methods is often constrained by computational needs, especially in modern contexts involving thousands of variables. To overcome this issue, we introduce two novel Markov chain Monte Carlo (MCMC) search algorithms that have a significantly lower computational cost than leading Bayesian approaches. Our proposed MCMC-based search algorithms use the marginal pseudo-likelihood approach to bypass the complexities of computing intractable normalizing constants and iterative precision matrix sampling. These algorithms can deliver reliable results in mere minutes on standard computers, even for large-scale problems with one thousand variables. Furthermore, our proposed method is capable of addressing model uncertainty by efficiently exploring the full posterior graph space. Our simulation study indicates that the proposed algorithms, particularly for large-scale sparse graphs, outperform the leading Bayesian approaches in terms of computational efficiency and precision. The implementation supporting the new approach is available through the R package BDgraph.Keywords
All Related Versions
This publication has 34 references indexed in Scilit:
- Bayesian Structure Learning in Undirected Gaussian Graphical Models: Literature Review with Empirical ComparisonJournal of the American Statistical Association, 2024
- Covariate-Assisted Bayesian Graph Learning for Heterogeneous DataJournal of the American Statistical Association, 2023
- TheG-Wishart Weighted Proposal Algorithm: Efficient Posterior Computation for Gaussian Graphical ModelsJournal of Computational and Graphical Statistics, 2022
- Accelerating Bayesian Structure Learning in Sparse Gaussian Graphical ModelsJournal of the American Statistical Association, 2021
- Informed Proposals for Local MCMC in Discrete SpacesJournal of the American Statistical Association, 2019
- Learning Gaussian graphical models with fractional marginal pseudo-likelihoodInternational Journal of Approximate Reasoning, 2017
- Bayesian Inference of Multiple Gaussian Graphical ModelsJournal of the American Statistical Association, 2015
- Efficient sampling of Gaussian graphical models using conditional Bayes factorsStat, 2014
- A direct sampler for G‐Wishart variatesStat, 2013
- A Monte Carlo method for computing the marginal likelihood in nondecomposable Gaussian graphical modelsBiometrika, 2005