Early Career Bayes Seminar for 2022

The Statistical Society of Australia (QLD Branch) and Centre for Data Science, Queensland University of Technology are delighted to have our second Early Career Bayes Seminar for 2022 presented by Mary Llewellyn from the School of Mathematics at the University of Edinburgh.

Virtual attendees: A Zoom link will be emailed prior to event.

Location: QUT Gardens Point, room TBA

Title: Discretising a Continuous World: Accelerated Inference for State-Space Models via Hidden Markov Models

Abstract: Inference on the process parameters of a state-space model (SSM) can be especially challenging when the likelihood of the data given the parameters is not available in closed form. A variety of approaches to Bayesian model fitting have been applied in such situations, including MCMC with data augmentation, sequential Monte Carlo (SMC) approximation, and particle MCMC algorithms. However, these existing methods can be inefficient because of sample impoverishment in the SMC approximations and/or poor mixing in the MCMC steps. This seminar will detail an approach that borrows ideas from discrete hidden Markov models (HMMs) to provide an efficient MCMC with data augmentation approach, imputing the latent states within the algorithm. Our approach deterministically approximates the SSM by a discrete HMM, which is subsequently used as an MCMC proposal distribution for the latent states. We demonstrate that the algorithm provides an efficient alternative approach via two different case studies.

To register please click here.