Tutorial on Sequential Monte Carlo methods in Statistics

Thanks to ACEMS for their support in for Anthony's visit.

Speaker: Dr Anthony Lee, University of Bristol

I will introduce Sequential Monte Carlo (SMC) methodology from a Statistics perspective. This particle-based algorithm was initially proposed to approximate predictive and filtering distributions for general state-space hidden Markov models, and in this context it is also known as a Particle Filter. SMC is now used to approximate a variety of intractable integrals arising in Statistics, e.g. intractable likelihoods in latent variable models and expectations with respect to high-dimensional, complex posteriors. I will cover the basic algorithm and its properties, as well as some innovations that have improved its performance and extended its impact. No previous knowledge of SMC is required.

Biography:

I am a Computational Statistician in the School of Mathematics at the University of Bristol, and Director of the Data Science at Scale Programme at the Alan Turing Institute, with Intel as Strategic Partner. My research is primarily in the area of stochastic algorithms for approximating intractable quantities that arise in data analysis. Examples of such algorithms are Markov chain and Sequential Monte Carlo. I work on both theory and methodology: research in this area is interdisciplinary, bringing together advances in applied probability, algorithms, and statistics. I am often interested in algorithms that scale well in parallel and distributed computing environments, and in computational and statistical trade-offs when conducting inference.