BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Online temporally adaptive parameter estimation with applications 
 to streaming data analysis. - Cristoforos Anagnostopoulos\, Statistics Lab
 oratory\, University of Cambridge
DTSTART:20101103T141500Z
DTEND:20101103T150000Z
UID:TALK27516@talks.cam.ac.uk
CONTACT:Rachel Fogg
DESCRIPTION:Online learning algorithms deployed in streaming data \ncontex
 ts may be additionally required to possess temporally adaptive properties\
 , in order to remain up-to-date against unforeseen changes\, smooth or abr
 upt\, in the underlying data generation mechanism. In cases where explicit
  dynamic modelling is either impossible or impractical\, temporally adapti
 ve behaviour may still be induced by controlling the responsiveness of the
  estimator to novel information.\nThis can be naturally accomplished in ce
 rtain algorithms that feature user-specified learning rates\, such as the 
 Robbins-Monro family of \nalgorithms. We discuss available methodology for
  automatic self-tuning learning rates in a Robbins-Monro context. On the b
 asis of both \ntheoretical insights and real-data experiments\, we demonst
 rate that this approach can efficiently handle temporal variation of unkno
 wn characteristics\, while additionally serving as a monitoring tool.\n
LOCATION:LR10\, Engineering\, Department of
END:VEVENT
END:VCALENDAR