Output details
15 - General Engineering
University of Oxford
Sequential non-stationary dynamic classification with sparse feedback
This paper developed a tractable methodology for streaming decision making in non-stationary environments, allowing missing observations to be imputed and label bit errors to be decoded and corrected as well as active observation requesting. This paper spawns from work funded by the UK EPSRC and was subsequently integrated into an EPSRC/industry platform grant "Aladdin" (won the 2009 Engineer award for best industry/academia project). The work on active observation requesting formed a major component in the formulation of a follow-on EPSRC/industry platform grant (project "Orchid"). Industry contact: BAE Systems (contact details available).