BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Inferring rapidly-varying big data time series--from deconvolution
  to a new form of time series model - Sofia Olhede\, UCL
DTSTART:20161130T140000Z
DTEND:20161130T150000Z
UID:TALK68887@talks.cam.ac.uk
CONTACT:Rachel Furner
DESCRIPTION:Traditional time series models can only encapsulate slow varia
 tion in the underlying generative mechanism. However\, in many scenarios\,
  this is not a realistic assumption. There seems to be an unavoidable conf
 lict between how rapidly the structured part of the model can change\, ver
 sus how much we need to average in order to retrieve parameters stably. We
  here introduce a new class of nonstationary time series\, and show how ef
 ficient and rapid inference is still possible in this scenario\, despite t
 he generating mechanism changing quickly. The methods are illustrated on d
 rifter time series\, from the global drifter programme. Computational effi
 ciency becomes a key constraint when handling  20000 long time series to o
 btain a global understanding of circulation\, making this a big data probl
 em. Depending on the latitude of the observations\, the underlying generat
 ive mechanism of the observed phenomenon is either slowly or rapidly chang
 ing\, and we show how the newly introduced methodology can resolve both sc
 enarios.\n\nThis is joint work with Arthur Guillaimin\, Adam Sykulski\, Je
 ffrey Early and Jonathan Lilly
LOCATION:MR5 Centre for Mathematical Sciences
END:VEVENT
END:VCALENDAR