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:Statistics
SUMMARY:The Bouncy Particle Sampler - Alexandre Bouchard-C
ôté (UBC)
DTSTART;TZID=Europe/London:20170505T160000
DTEND;TZID=Europe/London:20170505T170000
UID:TALK71958AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/71958
DESCRIPTION:Markov chain Monte Carlo methods have become stand
ard tools to sample from complex high-dimensional
probability measures. Many available techniques re
ly on discrete-time reversible Markov chains whose
transition kernels built up over the Metropolis-H
astings algorithm. In our recent work\, we investi
gate an alternative approach\, the Bouncy Particle
Sampler (BPS) where the target distribution of in
terest is explored using a continuous-time\, non r
eversible Markov process. In this alternative appr
oach\, a particle moves along straight lines conti
nuously around the space and\, when facing a high
energy barrier\, it is not rejected but its path i
s modified by bouncing against this barrier. The r
esulting non-reversible Markov process provides a
rejection-free Markov chain Monte Carlo sampling s
cheme. This method\, inspired from recent work in
the molecular simulation literature\, is shown to
be a valid\, efficient sampling scheme applicable
to a wide range of Bayesian problems. We present s
everal additional original methodological extensio
ns and establish various theoretical properties of
these procedures. We demonstrate experimentally t
he efficiency of these algorithms on a variety of
Bayesian inference problems.
LOCATION:MR12\, Centre for Mathematical Sciences\, Wilberfo
rce Road\, Cambridge.
CONTACT:Quentin Berthet
END:VEVENT
END:VCALENDAR