BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Signal Processing and Communications Lab Seminars
SUMMARY:Approximate Smoothing and Parameter Estimation in
High-Dimensional State-Space Models - Dr Axel Fink
e\, CUED
DTSTART;TZID=Europe/London:20160519T150000
DTEND;TZID=Europe/London:20160519T160000
UID:TALK65851AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/65851
DESCRIPTION:We present an approximate algorithm for estimating
additive smoothing functionals in a class of high
-dimensional state-space models via sequential Mon
te Carlo methods. In such high-dimensional setting
s\, a prohibitively large number of particles\, i.
e. growing exponentially in the dimension of the s
tate space\, is usually required to obtain useful
estimates of such smoothed quantities. Exploiting
spatial ergodicity properties of the model\, we ci
rcumvent this problem via a blocking strategy whic
h leads to approximations that can be computed rec
ursively in time and in parallel in space. In part
icular\, our method enables us to perform maximum-
likelihood estimation via stochastic gradient-asce
nt and stochastic EM algorithms. We demonstrate th
e method on a high-dimensional state-space model.
\n\nThis is joint work with Sumeetpal S. Singh.
LOCATION:LR6\, Department of Engineering
CONTACT:Dr Ramji Venkataramanan
END:VEVENT
END:VCALENDAR