Output details
15 - General Engineering
University of Oxford
Sequential Bayesian Prediction in the Presence of Changepoints and Faults
This paper developed, for the first time, a principled method based around Bayesian non-parametric models. The approach was refined and applied to sensor streams with faults, resulting in industrial takeup by BAE systems. This work was funded by an EPSRC/industry platform grant "Aladdin" (won the 2008&9 "Engineer" awards - best industry/academia project). It formed a foundational part of the "Orchid" grant, an £11m joint EPSRC/industry grant, a £1.2m EPSRC grant in "autonomous, intelligent systems", and a Defence Technology Centre project (QinetiQ, contact details available). Three joint-industry patents come from our work on Bayesian non-parametrics information fusion (WO 2012164243/4/5).