Asymptotics of cut distributions and robust modular inference using Posterior Bootstrap

Abstract

Bayesian inference provides a framework to combine various model components with shared parameters, allowing joint uncertainty estimation and the use of all available data sources. Unfortunately, misspecification of any part of the model might propagate to all other parts and can lead to unsatisfactory results. Cut distributions have been proposed as a remedy, where the information is prevented from flowing along certain directions. We study cut distributions from an asymptotic perspective and obtain a Bernstein-von Mises theorem, as well as a Laplace approximation with quantitative bounds. We then propose an algorithm based on the Posterior Bootstrap that delivers credible regions with the nominal frequentist asymptotic coverage. The proposed methods are illustrated with numerical experiments in a variety of examples, including causal inference with propensity scores.

Major revision, including new results on the control of the Laplace approximation error for cut posteriors

References in corpus (14)

• Practical Bayesian model evaluation using leave-one-out cross-validation and WAIC
• Better together? Statistical learning in models made of modules
• Sensitivity analysis via the proportion of unmeasured confounding
• Nonparametric learning from Bayesian models with randomized objective functions
• Scalable Nonparametric Sampling from Multimodal Posteriors with the Posterior Bootstrap
• Stochastic Approximation Cut Algorithm for Inference in Modularized Bayesian Models
• Semi-Modular Inference: enhanced learning in multi-modular models by tempering the influence of components
• Robust Inference and Model Criticism Using Bagged Posteriors
• Bayesian Bootstrap Spike-and-Slab LASSO
• Introducing prior information in Weighted Likelihood Bootstrap with applications to model misspecification
• A numerically stable algorithm for integrating Bayesian models using Markov melding
• Variational inference for cutting feedback in misspecified models
• On an Asymptotic Distribution for the MLE
• A conditional independence framework for coherent modularized inference