Scalable Bayesian Learning for State Space Models using Variational Inference with SMC Samplers
This work addresses the need for scalable Bayesian methods in complex models like stochastic volatility and point processes, offering an incremental improvement over existing variational EM approximations.
The authors tackled the problem of performing scalable fully Bayesian inference in state space models, presenting a method that combines variational inference with sequential Monte Carlo sampling to achieve efficient inference for both latent states and static parameters.
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.