BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Challenges in implementing the Bayesian paradigm - Prof. Steve Mac
 Eachern (Ohio State)
DTSTART:20110310T120000Z
DTEND:20110310T130000Z
UID:TALK30214@talks.cam.ac.uk
CONTACT:Zoubin Ghahramani
DESCRIPTION:The optimality properties of Bayesian inference are well estab
 lished\, and yet there remains a wide gulf between the mathematical founda
 tions of the methods and practical implementation.  The gap shows up most 
 strongly in specification of the prior distribution\, particularly when th
 e analyst has (or wishes to inject) little information about the parameter
 s.  In such cases\, the prior distribution often depends on the data itsel
 f\, and this dependence is routinely ignored in the subsequent analysis.  
 In this talk\, we formalize the notion of a data dependent prior distribut
 ion and show how to bring the analysis into agreement with Bayes Theorem. 
  Properties of the adjustment are described\, and a range of impacts on po
 sterior inference is illustrated.  The impact is particularly large in hig
 h and infinite dimensional settings\, such as those characterizing nonpara
 metric Bayesian inference.  In this high-dimensional setting\, a quick des
 cription of the need for additional adjustments to traditional Bayesian co
 ncepts such as the Bayes factor will be given.  \n\nThe work on data depen
 dent prior distributions is joint with Bill Darnieder\; that on Bayes fact
 ors is joint with Xinyi Xu\, Pingbo Lu\, and Ruoxi Xu.  \n
LOCATION:Engineering Department\, CBL Room 438
END:VEVENT
END:VCALENDAR