BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Nonparametric Bayesian statistics with exchangeable random structu
 res - Daniel Roy
DTSTART:20131011T110000Z
DTEND:20131011T140000Z
UID:TALK48261@talks.cam.ac.uk
CONTACT:Gintare Karolina Dziugaite
DESCRIPTION:Most of nonparametric Bayesian statistics is focused on the se
 ttings of "i.i.d. data" and "regression with i.i.d. noise".  What of probl
 ems that don't fit into one of these molds?  I'll introduce exchangeabilit
 y (and other invariance principles) as a general guiding principle for con
 structing statistical models\, and in particular identifying appropriate p
 arameter spaces.  The main focus will be on networks and graphs\, where ex
 changeability of vertices is shown by Aldous-Hoover to give a natural para
 meter space of "graphons"\, i.e.\, measurable functions from [0\,1]^2 to [
 0\,1].  I'll give a few more examples of exchangeability\, including Marko
 v exchangeability and rotatability.  I'll start with the sequence case to 
 explain how to interpret/understand the later theorems.
LOCATION:Engineering Department\, CBL Room BE-438
END:VEVENT
END:VCALENDAR