Speeding Up MCMC by Delayed Acceptance and Data Subsampling

Abstract

The complexity of the Metropolis-Hastings (MH) algorithm arises from the requirement of a likelihood evaluation for the full data set in each iteration. Payne and Mallick (2015) propose to speed up the algorithm by a delayed acceptance approach where the acceptance decision proceeds in two stages. In the first stage, an estimate of the likelihood based on a random subsample determines if it is likely that the draw will be accepted and, if so, the second stage uses the full data likelihood to decide upon final acceptance. Evaluating the full data likelihood is thus avoided for draws that are unlikely to be accepted. We propose a more precise likelihood estimator which incorporates auxiliary information about the full data likelihood while only operating on a sparse set of the data. We prove that the resulting delayed acceptance MH is more efficient compared to that of Payne and Mallick (2015). The caveat of this approach is that the full data set needs to be evaluated in the second stage. We therefore propose to substitute this evaluation by an estimate and construct a state-dependent approximation thereof to use in the first stage. This results in an algorithm that (i) can use a smaller subsample m by leveraging on recent advances in Pseudo-Marginal MH (PMMH) and (ii) is provably within $O(m^{-2})$ of the true posterior.

Accepted for publication in Journal of Computational and Graphical Statistics

References in corpus (6)

• The pseudo-marginal approach for efficient Monte Carlo computations
• On Markov chain Monte Carlo methods for tall data
• Firefly Monte Carlo: Exact MCMC with Subsets of Data
• Coupled MCMC with a randomized acceptance probability
• Accelerating Metropolis-Hastings algorithms: Delayed acceptance with prefetching
• Light and Widely Applicable MCMC: Approximate Bayesian Inference for Large Datasets

Cited by in corpus (14)

• Approximating Bayes in the 21st Century
• Hamiltonian Monte Carlo with Energy Conserving Subsampling
• The Block Pseudo-Marginal Sampler
• Importance sampling correction versus standard averages of reversible MCMCs in terms of the asymptotic variance
• Computing Bayes: From Then 'Til Now'
• Accelerating delayed-acceptance Markov chain Monte Carlo algorithms
• A survey of Monte Carlo methods for noisy and costly densities with application to reinforcement learning and ABC
• Spectral Subsampling MCMC for Stationary Time Series
• Flexible Bayesian Nonlinear Model Configuration
• The block-Poisson estimator for optimally tuned exact subsampling MCMC
• Deep Bayesian regression models
• Accelerating sequential Monte Carlo with surrogate likelihoods
• Scalable Metropolis-Hastings for Exact Bayesian Inference with Large Datasets
• Subsampling MCMC - An introduction for the survey statistician