MCMC for non-linear state space models using ensembles of latent sequences

TL;DR

This paper introduces a novel MCMC method for non-linear state space models that uses large ensembles of latent sequences, significantly improving inference efficiency over existing methods, demonstrated on a population dynamics model.

Contribution

The paper proposes a new ensemble-based MCMC technique for non-linear state space models, enhancing sampling efficiency by using large latent sequence ensembles and partial backward passes.

Findings

Ensemble method is 1.9 to 12 times more efficient than single-sequence embedded HMM.

Ensemble method outperforms simple Metropolis sampling.

Partial backward passes further increase efficiency by 3.4 to 20.4 times.

Abstract

Non-linear state space models are a widely-used class of models for biological, economic, and physical processes. Fitting these models to observed data is a difficult inference problem that has no straightforward solution. We take a Bayesian approach to the inference of unknown parameters of a non-linear state model; this, in turn, requires the availability of efficient Markov Chain Monte Carlo (MCMC) sampling methods for the latent (hidden) variables and model parameters. Using the ensemble technique of Neal (2010) and the embedded HMM technique of Neal (2003), we introduce a new Markov Chain Monte Carlo method for non-linear state space models. The key idea is to perform parameter updates conditional on an enormously large ensemble of latent sequences, as opposed to a single sequence, as with existing methods. We look at the performance of this ensemble method when doing Bayesian inference in the Ricker model of population dynamics. We show that for this problem, the ensemble method is vastly more efficient than a simple Metropolis method, as well as 1.9 to 12.0 times more efficient than a single-sequence embedded HMM method, when all methods are tuned appropriately. We also introduce a way of speeding up the ensemble method by performing partial backward passes to discard poor proposals at low computational cost, resulting in a final efficiency gain of 3.4 to 20.4 times over the single-sequence method.