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:Isaac Newton Institute Seminar Series
SUMMARY:Least-action filtering - Rogers\, C (Cambridge)
DTSTART;TZID=Europe/London:20100615T095000
DTEND;TZID=Europe/London:20100615T104000
UID:TALK25292AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/25292
DESCRIPTION:This talk studies the filtering of a partially-obs
erved multidimensional diffusion process using the
principle of least action\, equivalently\, maximu
m-likelihood estimation. We show how the most like
ly path of the unobserved part of the diffusion ca
n be determined by solving a shooting ODE\, and th
en we go on to study the (approximate) conditional
distribution of the diffusion around the most lik
ely path\; this turns out to be a zero-mean Gaussi
an process which solves a linear SDE whose time-de
pendent coefficients can be identified by solving
a first-order ODE with an initial condition. This
calculation of the conditional distribution can be
used as a way to guide SMC methods to search rele
vant parts of the state space\, which may be valua
ble in high-dimensional problems\, where SMC strug
gles\; in contrast\, ODE solution methods continue
to work well even in moderately large dimension.
LOCATION:Seminar Room 1\, Newton Institute
CONTACT:Mustapha Amrani
END:VEVENT
END:VCALENDAR