Stochastic Bouncy Particle Sampler

Abstract

We introduce a novel stochastic version of the non-reversible, rejection-free Bouncy Particle Sampler (BPS), a Markov process whose sample trajectories are piecewise linear. The algorithm is based on simulating first arrival times in a doubly stochastic Poisson process using the thinning method, and allows efficient sampling of Bayesian posteriors in big datasets. We prove that in the BPS no bias is introduced by noisy evaluations of the log-likelihood gradient. On the other hand, we argue that efficiency considerations favor a small, controllable bias in the construction of the thinning proposals, in exchange for faster mixing. We introduce a simple regression-based proposal intensity for the thinning method that controls this trade-off. We illustrate the algorithm in several examples in which it outperforms both unbiased, but slowly mixing stochastic versions of BPS, as well as biased stochastic gradient-based samplers.

ICML Camera ready version

Cited by in corpus (12)

• Piecewise-Deterministic Markov Chain Monte Carlo
• Ergodicity of the zigzag process
• Piecewise Deterministic Markov Processes for Scalable Monte Carlo on Restricted Domains
• PASS-GLM: polynomial approximate sufficient statistics for scalable Bayesian GLM inference
• Binary Bouncy Particle Sampler
• The Coordinate Sampler: A Non-Reversible Gibbs-like MCMC Sampler
• Non-reversible, tuning- and rejection-free Markov chain Monte Carlo via iterated random functions
• Analysis of high-dimensional Continuous Time Markov Chains using the Local Bouncy Particle Sampler
• Bouncy Hybrid Sampler as a Unifying Device
• Generalizing Parallel Replica Dynamics: Trajectory Fragments, Asynchronous Computing, and PDMPs
• A stepped sampling method for video detection using LSTM
• New Heuristics for Parallel and Scalable Bayesian Optimization