Spectral Subsampling MCMC for Stationary Time Series
arXiv:1910.13627
Abstract
Bayesian inference using Markov Chain Monte Carlo (MCMC) on large datasets has developed rapidly in recent years. However, the underlying methods are generally limited to relatively simple settings where the data have specific forms of independence. We propose a novel technique for speeding up MCMC for time series data by efficient data subsampling in the frequency domain. For several challenging time series models, we demonstrate a speedup of up to two orders of magnitude while incurring negligible bias compared to MCMC on the full dataset. We also propose alternative control variates for variance reduction based on data grouping and coreset constructions.
Empirical section significantly revised and extended
References in corpus (10)
- The pseudo-marginal approach for efficient Monte Carlo computations
- Stochastic Gradient Hamiltonian Monte Carlo
- Speeding Up MCMC by Efficient Data Subsampling
- Central limit theorem for Fourier transforms of stationary processes
- Automated Scalable Bayesian Inference via Hilbert Coresets
- Speeding Up MCMC by Delayed Acceptance and Data Subsampling
- Bayesian nonparametric spectral density estimation using B-spline priors
- The Block Pseudo-Marginal Sampler
- The block-Poisson estimator for optimally tuned exact subsampling MCMC
- Central limit theorem for Fourier transform and periodogram of random fields