Scalable Bayesian Learning for State Space Models using Variational Inference with SMC Samplers

TL;DR

This paper introduces a scalable Bayesian inference method for state space models that combines variational inference with SMC samplers, enabling full Bayesian analysis of latent states and parameters.

Contribution

It proposes a novel approach that integrates variational methods with SMC samplers, outperforming particle MCMC in scalability and full Bayesian inference capabilities.

Findings

Enables scalable inference in multivariate stochastic volatility models

Allows flexible dynamics in latent intensity functions of point processes

Demonstrates advantages over variational EM in static parameter inference

Abstract

We present a scalable approach to performing approximate fully Bayesian inference in generic state space models. The proposed method is an alternative to particle MCMC that provides fully Bayesian inference of both the dynamic latent states and the static parameters of the model. We build up on recent advances in computational statistics that combine variational methods with sequential Monte Carlo sampling and we demonstrate the advantages of performing full Bayesian inference over the static parameters rather than just performing variational EM approximations. We illustrate how our approach enables scalable inference in multivariate stochastic volatility models and self-exciting point process models that allow for flexible dynamics in the latent intensity function.