BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Comparing Consensus Monte Carlo Strategies for Distributed Bayesia
 n Computation - Steven Scott (Google)
DTSTART:20170704T100000Z
DTEND:20170704T104500Z
UID:TALK73140@talks.cam.ac.uk
CONTACT:INI IT
DESCRIPTION:Consensus Monte Carlo is an algorithm for conducting Monte Car
 lo  based Bayesian inference on large data sets distributed across many  w
 orker machines in a data center.  The algorithm operates by running  a sep
 arate Monte Carlo algorithm on each worker machine\, which only  sees a po
 rtion of the full data set.  The worker-level posterior  samples are then 
 combined to form a Monte Carlo approximation to the  full posterior distri
 bution based on the complete data set.  We  compare several methods of car
 rying out the combination\, including a  new method based on approximating
  worker-level simulations using a  mixture of multivariate Gaussian distri
 butions.  We find that  resampling and kernel density based methods break 
 down after 10 or  sometimes fewer dimensions\, while the new mixture-based
  approach  works well\, but the necessary mixture models take too long to 
 fit.
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR