BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Virtual Seminar: ‘Score driven modeling of  spatio-temporal data
 ’ - Alessandra Luati\, University of Bologna 
DTSTART:20201119T140000Z
DTEND:20201119T150000Z
UID:TALK153883@talks.cam.ac.uk
CONTACT:Alison Quenault
DESCRIPTION:A simultaneous autoregressive score driven model is developed 
 for spatio-temporal data that are generated by a multivariate Student-t di
 stribution. The model specification rests on a signal plus noise decomposi
 tion of a spatially filtered process\, where the signal can be approximate
 d by a non linear function of the past variables and the noise follows a S
 tudent-t distribution. The key feature of the model is that the dynamics o
 f the space-time  varying signal are driven by the score of the conditiona
 l likelihood function.  When the distribution is heavy tailed\, the score 
 provides a robust update of the space-time varying location. Consistency a
 nd asymptotic normality of maximum likelihood estimators are derived along
  with the stochastic properties of the model. The motivating application o
 f the proposed model comes from brain scans recorded through functional ma
 gnetic resonance imaging when subjects are at rest and not expected to rea
 ct to any controlled stimulus. We identify spontaneous activations in brai
 n regions as extreme values of a possibly heavy tailed distribution\, by a
 ccounting for spatial and temporal dependence. \n\nJoint work with: France
 sca Gasperoni (University of Cambridge)\, Lucia Paci (University of Milano
  CSC)\, Enzo D’Innocenzo (University of Bologna)\n
LOCATION:Virtual Seminar 
END:VEVENT
END:VCALENDAR