Importance sampling type estimators based on approximate marginal MCMC

Abstract

We consider importance sampling (IS) type weighted estimators based on Markov chain Monte Carlo (MCMC) targeting an approximate marginal of the target distribution. In the context of Bayesian latent variable models, the MCMC typically operates on the hyperparameters, and the subsequent weighting may be based on IS or sequential Monte Carlo (SMC), but allows for multilevel techniques as well. The IS approach provides a natural alternative to delayed acceptance (DA) pseudo-marginal/particle MCMC, and has many advantages over DA, including a straightforward parallelisation and additional flexibility in MCMC implementation. We detail minimal conditions which ensure strong consistency of the suggested estimators, and provide central limit theorems with expressions for asymptotic variances. We demonstrate how our method can make use of SMC in the state space models context, using Laplace approximations and time-discretised diffusions. Our experimental results are promising and show that the IS type approach can provide substantial gains relative to an analogous DA scheme, and is often competitive even without parallelisation.

34 pages, 1 figure

References in corpus (7)

• The pseudo-marginal approach for efficient Monte Carlo computations
• Three discussions of the paper "sequential quasi-Monte Carlo sampling", by M. Gerber and N. Chopin
• Harris recurrence of Metropolis-within-Gibbs and trans-dimensional Markov chains
• Delayed acceptance particle MCMC for exact inference in stochastic kinetic models
• Uniform ergodicity of the Particle Gibbs sampler
• An Introduction to Twisted Particle Filters and Parameter Estimation in Non-linear State-space Models
• Importance sampling correction versus standard averages of reversible MCMCs in terms of the asymptotic variance

Cited by in corpus (5)

• Importance sampling correction versus standard averages of reversible MCMCs in terms of the asymptotic variance
• bssm: Bayesian Inference of Non-linear and Non-Gaussian State Space Models in R
• Sampling algorithms in statistical physics: a guide for statistics and machine learning
• Efficient Bayesian generalized linear models with time-varying coefficients: The walker package in R
• Simulating counterfactuals