Dr Matt Sutton, Imke Botha and Dr Joshua Bon
Queensland University of Technology
An Introduction to Sequential Monte Carlo
Sequential Monte Carlo (SMC) is a versatile algorithmic tool for data science, machine learning and statistics. It can be used for a myriad of inferential problems dealing with latent state prediction and parameter estimation. SMC Samplers provide robust parameter inference with uncertainty quantification for static Bayesian models. Whilst Particle filters, a subset of SMC algorithms, are popular for efficient estimation of latent states in hidden Markov models, beyond the limitations of the Kalman filter. This one-day introduction will orientate you to the world of SMC, demystify notation, and provide you with some hands on coding experience with SMC in Julia.
Topics:
- An introduction to SMC Samplers (Matt Sutton)
- An introduction to Particle Filters (Imke Botha)
- An introduction to general SMC (Joshua Bon)
Pre-requisites
This course assumes an undergraduate level understanding of statistics. Coding will use Julia, with a focus on understanding the template code and making adjustments.
Dr Matt Sutton
Queensland University of Technology
Dr Sutton is a postdoc researcher with a strong interest in developing new methods and algorithms for modern statistical challenges. He is particularly interested in Bayesian statistics and computational challenges involving: working with big data, model averaging, and conducting inference with privacy constraints.
Imke Botha
Queensland University of Technology
Imke Botha is a PhD candidate at the Queensland University of Technology. Her research interests include sequential Monte Carlo, Markov chain Monte Carlo and pseudo-marginal methods. Her current research focuses on Bayesian inference for intractable likelihood models.
Dr Joshua Bon
Queensland University of Technology
Dr Joshua Bon is a Research Fellow in sequential Monte Carlo at the Queensland University of Technology. His research involves developing efficient and robust algorithms for Bayesian inference along with their mathematical justifications.