BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Exact Bayesian inference for change point models with application 
 to genomics - Robin\, S (INRA - Institut National de la Recherche Agronomi
 que)
DTSTART:20140203T140000Z
DTEND:20140203T150000Z
UID:TALK50670@talks.cam.ac.uk
CONTACT:Mustapha Amrani
DESCRIPTION:We are interested in the evaluation of uncertainty in the loca
 tion of change points. Because of the discrete nature of change points\, s
 tandard statistical theory does not apply easily and only few asymptotic r
 esults are available to evaluate quantity of interest such as confidence i
 ntervals for change point locations. In a Bayesian framework\, such quanti
 ties are often obtained via computationally demanding Monte-Carlo techniqu
 es. We will present a general Bayesian methodology to compute such quantit
 ies in an exact manner\, with quadratic complexity. A parallel can be made
  between this approach and dynamic programming algorithms. This methodolog
 y allows to compute a series of posterior distributions\, such as this of 
 the total number of breakpoints or this of any change point location. Base
 d on these results we will consider a Bayesian method to compare the locat
 ion of change points in independent sequences. Eventually\, this method wi
 ll be used to compare the location of transcript boundaries in yeast under
  different growth condition.\n
LOCATION:Seminar Room 2\, Newton Institute Gatehouse
END:VEVENT
END:VCALENDAR