Data augmentation for models based on rejection sampling

TL;DR

This paper introduces a data augmentation method for Markov chain Monte Carlo inference in models involving rejection sampling, simplifying complex distributions and improving sampling efficiency across various challenging problems.

Contribution

The authors propose a novel data augmentation scheme that explicitly models rejected proposals to facilitate inference in intractable models involving rejection sampling.

Findings

Outperforms existing sampling algorithms in experiments

Effective in modeling truncated flow-cytometry data

Improves Bayesian inference for doubly-intractable problems

Abstract

We present a data augmentation scheme to perform Markov chain Monte Carlo inference for models where data generation involves a rejection sampling algorithm. Our idea, which seems to be missing in the literature, is a simple scheme to instantiate the rejected proposals preceding each data point. The resulting joint probability over observed and rejected variables can be much simpler than the marginal distribution over the observed variables, which often involves intractable integrals. We consider three problems, the first being the modeling of flow-cytometry measurements subject to truncation. The second is a Bayesian analysis of the matrix Langevin distribution on the Stiefel manifold, and the third, Bayesian inference for a nonparametric Gaussian process density model. The latter two are instances of problems where Markov chain Monte Carlo inference is doubly-intractable. Our experiments demonstrate superior performance over state-of-the-art sampling algorithms for such problems.