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:SMC Samplers for Applications in High Dimensions -
Dr Alexandros Beskos\, University College London
DTSTART;TZID=Europe/London:20141127T141500
DTEND;TZID=Europe/London:20141127T151500
UID:TALK55633AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/55633
DESCRIPTION:Sequential Monte Carlo (SMC) methods are nowadays
routinely applied in a variety of complex applicat
ions: hidden Markov models\, dynamical systems\, t
arget tracking\, control problems\, just to name a
few. Whereas SMC methods have been greatly refine
d in the last decades and are now much better unde
rstood\, they are still known to suffer from the c
urse of dimensionality: algorithms can sometimes b
reak down exponentially fast with the dimension of
the state space. As a consequence\, practitioners
in high-dimensional Data Assimilation application
s in atmospheric sciences\, oceanography and elsew
here will typically use 3D-Var or Kalman-filter-ty
pe approximations that will provide biased estimat
es in the presence of non-linear model dynamics.\n
\nThe talk will concentrate on a class of SMC algo
rithms and will look at ways to reduce the cost of
the algorithms as a function of the dimension of
the state space. Explicit asymptotic results will
clarify the effect of the dimension at the propert
ies of the algorithm and could provide a platform
for algorithmic optimisation in high dimensions. A
pplications will be shown in the context of Data A
ssimilation\, in a problem where the objective is
to target the posterior distribution of the initia
l condition of the Navier-Stokes equation given a
Gaussian prior and noisy observations at different
instances and locations of the spatial field. The
dimension of the signal is in theory infinite-dim
ensional - in practice 64x64 or more depending on
the resolution – thus posing great challenges for
the development and efficiency of SMC methods.
LOCATION:Board Room\, CUED
CONTACT:Fredrik Lindsten
END:VEVENT
END:VCALENDAR