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:Machine Learning @ CUED
SUMMARY:Nonparametric Bayesian statistics with exchangeabl
e random structures - Daniel Roy
DTSTART;TZID=Europe/London:20131011T120000
DTEND;TZID=Europe/London:20131011T150000
UID:TALK48261AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/48261
DESCRIPTION:Most of nonparametric Bayesian statistics is focus
ed on the settings of "i.i.d. data" and "regressio
n with i.i.d. noise". What of problems that don't
fit into one of these molds? I'll introduce exch
angeability (and other invariance principles) as a
general guiding principle for constructing statis
tical models\, and in particular identifying appro
priate parameter spaces. The main focus will be o
n networks and graphs\, where exchangeability of v
ertices is shown by Aldous-Hoover to give a natura
l parameter space of "graphons"\, i.e.\, measurabl
e functions from [0\,1]^2 to [0\,1]. I'll give a
few more examples of exchangeability\, including M
arkov exchangeability and rotatability. I'll star
t with the sequence case to explain how to interpr
et/understand the later theorems.
LOCATION:Engineering Department\, CBL Room BE-438
CONTACT:Gintare Karolina Dziugaite
END:VEVENT
END:VCALENDAR