James Murphy

Network structure inference

Part of my work during my PhD has been on statistical inference of network structure.  This is an important problem in a number of areas, including in tracking (where I have focussed my attention for now), systems biology, finance and sociology.  My method, which, as far as I know is new, allows the inference of the functional form of inter-object relationships to be deduced from observations of the objects in the network over time.  In its present form, the method requires fairly slow batch processing of observations in order to derive the relationships, but I think the work points the way towards future sequential inference methods for this problem, including with noisy observations of network components.  This is an active area of my research.  I have had a paper on this topic accepted into the FUSION international conference on data fusion, to be held in Singapore in July 2012.  I will link to this here when published.