Bayesian Analysis of High-dimensional Discrete Graphical Models

Abstract

This work introduces a Bayesian methodology for fitting large discrete graphical models with spike-and-slab priors to encode sparsity. We consider a quasi-likelihood approach that enables node-wise parallel computation resulting in reduced computational complexity. We introduce a scalable Langevin MCMC algorithm for sampling from the quasi-posterior distribution which enables variable selection and estimation simultaneously. We present extensive simulation results to demonstrate scalability and accuracy of the method. We also analyze the 16 Personality Factors (PF) dataset to illustrate performance of the method.

References in corpus (11)

• Improved contact prediction in proteins: Using pseudolikelihoods to infer Potts models
• An MCMC Approach to Classical Estimation
• High-dimensional Ising model selection using ${\ell }_1$-regularized logistic regression
• Estimating psychopathological networks: be careful what you wish for
• MCMC for doubly-intractable distributions
• Bayesian variable selection with shrinking and diffusing priors
• On Russian Roulette Estimates for Bayesian Inference with Doubly-Intractable Likelihoods
• Quasi-Bayesian analysis of nonparametric instrumental variables models
• Change-Point Estimation in High-Dimensional Markov Random Field Models
• Estimating heterogeneous graphical models for discrete data with an application to roll call voting
• An approach to large-scale Quasi-Bayesian inference with spike-and-slab priors