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:Semantics Lunch (Computer Laboratory)
SUMMARY:A Model-Learner Pattern for Bayesian Reasoning - A
ndy Gordon\, Microsoft Research and University of
Edinburgh
DTSTART;TZID=Europe/London:20121119T130000
DTEND;TZID=Europe/London:20121119T140000
UID:TALK41245AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/41245
DESCRIPTION:A Bayesian model consists of a pair of probability
distributions\, known as the prior and sampling d
istributions. A wide range of fundamental machine
learning tasks\, including regression\, classifica
tion\, clustering\, and many others\, can all be s
een as Bayesian models. We propose a new probabili
stic programming abstraction\, a typed Bayesian mo
del\, which is a pair of probabilistic functions f
or the prior and sampling distributions. A sampler
for a model is an algorithm to compute synthetic
data from its sampling distribution\, while a lear
ner for a model is an algorithm for probabilistic
inference on the model.\nModels\, samplers\, and l
earners form a generic programming pattern for mod
el-based inference.\nThey support the uniform expr
ession of common tasks including model testing\, a
nd generic compositions such as mixture models\, e
vidence-based model averaging\, and mixtures of ex
perts. A formal semantics supports reasoning about
model equivalence and implementation correctness.
By developing a series of examples and three lear
ner implementations based on exact inference\, bel
ief-propagation\, and Markov chain Monte Carlo\,\n
we demonstrate the broad applicability of this new
programming pattern.\n\nThe talk is based on join
t work with Mihhail Aizatulin (Open University)\,
Johannes Borgstroem (Uppsala University)\, Guillau
me Claret (MSR)\, Thore Graepel (MSR)\, Aditya Nor
i (MSR)\, Sriram Rajamani (MSR)\, and Claudio Russ
o (MSR).\n\nSee http://research.microsoft.com/fun
LOCATION:FW26
CONTACT:Peter Sewell
END:VEVENT
END:VCALENDAR