These information pages contain an introduction to my work and interests. The bulk of the articles relate to the research that I am currently undertaking.
Although by no means exhaustive, my research covers the following areas
Machine Learning Markets: Combining multiple subjective beliefs is an open research area in Bayesian theory and practice. By developing approaches for combining Bayesian beliefs, we can develop machine learning methods that can use the results of previously developed methods, and can work in conjunction with other methods. Markets provide a solid foundation for that approach.
Bayesian Methods for Brain Imaging: the development of Bayesian modelling methods in brain imaging, including diffusion MRI and functional MRI.
Continuous Time/Depth Systems: Learning and Inference. Many real systems are best described in continuous time, e.g. by differential equations or jump processes. Methods for fitting these to data are in their infancy, but critical in many areas: climate models, environmental models, brain imaging models, phylogenetics.
Dynamical Boltzmann Machine Models: Developing generic machine learning tools for sequences that can capture structured information about the temporal system.
Scalable Deep Learning: How to learn structured representations for systems in an inherently parallelisable and local way.
Other areas of work that relate to this include methods for modelling and understanding images, encorporating new machine learning knowledge into computational neuroscientific models, sensor networks, machine learning for astronomy, machine learning in changing environments and more...