Reading group

We meet weekly for a Bayesian computation reading group to discuss recent and classic papers.

Current group members

Mark Bell is a postdoc on my EPSRC project "Tractable inference for statistical network models with local dependence”. He is working on improving Bayesian inference techniques for statistical models for networks.

Richard Culliford is a PhD student funded by the University of Reading and the Modernising Medical Microbiology consortium at the University of Oxford. He is working on using sequential Monte Carlo for online Bayesian inference for the coalescent and for Bayesian model comparison, collaborating with Daniel Wilson.

Ivis Kerama is a PhD student funded by the EPSRC Mathematics of Planet Earth CDT and Cefas. He is working on "Improved Approximate Bayesian Computation for inference of the likely effects of climate change on animal populations”, co-supervised by Richard Sibly and Robert Thorpe. This project is part of an initiative on Individual-Based Modelling at the University of Reading.

Laura Mansfield is a PhD student funded by the EPSRC Mathematics of Planet Earth CDT. She is working on "Model reduction using emulation for understanding and predicting climate responses to different regional emission forcing”, co-supervised by Brian Hoskins and Apostolos Voulgarakis.

Philip Maybank is an EPSRC PhD student working on statistical methods for neuroscience. He has focussed on new methods for parameter estimation for differential equation models of brain activity from EEG data, in collaboration with Ingo Bojak. Philip is blogging about his research here.

Felipe Medina Aguayo is a postdoc on my BBSRC project “Understanding recombination through tractable statistical analysis of whole genome sequences”. He is working on applying recently developed techniques in Bayesian computation to whole genome sequence data, and on developing his own Bayesian computation methodology.

Changqiong Wang is an EPSRC PhD student who was recently awarded her thesis entitled "Applications of Monte Carlo methods in studying polymer dynamics”. She has been using Bayesian statistics to fit statistical models to molecular dynamics data, and is co-supervised by Zuowei Wang and Patrick Ilg. Changqiong's research project was instigated by the brilliant physicist Alexei Likhtman, who tragically died several years ago.

Previous group members

Kevin Synnott was an MRes student funded by the EPSRC Mathematics of Planet Earth CDT, co-supervised by Heather Graven and Felipe Medina Aguayo, whose thesis was entitled "Inferring Gaussian Markov random field models for estimating the surface ocean dissolved CO2 distribution”.

Paulina Rowinska was an MRes student funded by the EPSRC Mathematics of Planet Earth CDTwhose thesis was entitled "Approximate Bayesian Computation for expensive simulators”, co-supervised by Richard Sibly and Elske van der Vaart. Paulina contributes to several blogs: herehere and here, and is also active on twitter.

Ellen Hermes was an MSc student whose thesis was entitled "Bayesian model selection for brain network models”. Ellen has a twitter account.

Melina Evdemon was an MSc student whose thesis was entitled “Approximate Bayesian Computation for Network Models”. She has a linkedin account.

Other projects

In recent years I have focussed on developing the methodology of sequential Monte Carlo (SMC) and approximate Bayesian computation (ABC), motivated by applications in several fields. Particular focuses have been
  • Bayesian inference for doubly intractable distributions, i.e. distributions with an intractable partition function such as Markov random field models (including exponential random graph models (ERGMs)).
  • "Noisy" Monte Carlo methods, which approximate exact methods.
  • Bayesian model comparison.
I was formerly a postdoc on the Modernising Medical Microbiology project at the University of Oxford, and before that a Brunel Fellow in Statistics under the SuSTaIn programme in the Department of Mathematics at the University of Bristol (where I also studied for my PhD under the supervision of Peter Green). Prior to this I was a researcher at QinetiQ, specialising in target tracking, signal and image processing and classification.

Links to all of my publications, and to my PhD thesis, can be found on my Google Scholar profile