BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:The Correlated Pseudo-Marginal Method for Inference in Latent Vari
 able Models
DTSTART:20161017T171500
DTEND:20161017T183000
DTSTAMP:20260601T064316Z
UID:9429985bc567ba619e3a32958f83f5863ca5a9240ca5ec4e6a404703
CATEGORIES:Conferences - Seminars
DESCRIPTION:Prof. Arnaud Doucet\, University of Oxford\nThe pseudo-margina
 l algorithm is a popular variant of the Metropolis--Hastings scheme which 
 allows us to sample asymptotically from a target probability density when 
 we are only able to estimate an unnormalized version of this target unbias
 edly. It has found numerous applications in Bayesian statistics as there a
 re many scenarios where the likelihood function is intractable but can be 
 estimated unbiasedly using Monte Carlo samples. For a fixed computing time
 \, it has been shown in several recent contributions that an efficient imp
 lementation of the pseudo-marginal method requires the variance of the log
 -likelihood ratio estimator appearing in the acceptance probability of the
  algorithm to be of order 1\, which in turn requires scaling the number of
  Monte Carlo samples linearly with the number of data points. We propose a
  modification of the pseudo-marginal algorithm\, termed the correlated pse
 udo-marginal algorithm\, which is based on a novel log-likelihood ratio es
 timator computed using the difference of two positively correlated log-lik
 elihood estimators. This approach allows us to scale the number of Monte C
 arlo samples sub-linearly with the number of data points. A non-standard w
 eak convergence analysis of the method will be presented. In our numerical
  examples\, the efficiency of computations is increased relative to the ps
 eudo-marginal by up to several orders of magnitude for large datasets. Thi
 s is joint work George Deligiannidis and Michael K. Pitt: http://arxiv.org
 /abs/1511.04992
LOCATION:MA11 http://plan.epfl.ch/?lang=fr&room=cm01
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR