Data Assimilation Meetings at Reading

Date Meeting type Speakers
27 Feb 2013 Internal speaker Richard Everitt (UoR)
Recent developments in Monte Carlo methods (applied to Markov random fields) (pdf)

In this talk I will present some recently developed Monte Carlo algorithms, and briefly discuss some possible applications to data assimilation. My work on these methods has involved several different applications; this talk will mostly discuss Markov random field models (models involving many interacting variables), but many of the points made are not specific to these models. In general, in a range of applications, including population genetics, epidemic modelling and social network analysis, the data from which we wish to estimate parameters of interest consists of noisy or incomplete observations of an unobserved process. Bayesian statistics offers a framework in which to tackle this problem, accurately accounting for the uncertainty present due to the missing data. However, standard Markov chain Monte Carlo (MCMC) methods that are used to implement the Bayesian approach can perform poorly in this situation. In this talk I describe two alternatives to standard MCMC approaches: approximate Bayesian computation (ABC) and particle MCMC. Both methods are applied to parameter estimation of a hidden Markov random field, and are compared to the standard data augmentation approach.

Page navigation