BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:An Adaptive Sequential Monte Carlo Algorithm For Bayesian Mixture 
 Analysis - Ben Taylor\, Dept of Maths &amp\; Statistics\, Lancaster Univer
 sity
DTSTART:20091104T141500Z
DTEND:20091104T151500Z
UID:TALK19360@talks.cam.ac.uk
CONTACT:Rachel Fogg
DESCRIPTION:Particle filtering methodology is usually applied to non-linea
 r dynamic state space models\, but it is possible to adapt the methods to 
 work in static (ie fixed dataset) scenarios. This strategy is advantageous
  because it combines the best aspects of sequential importance sampling an
 d MCMC. By propagating a swarm of particles across the state space\, SMC a
 lgorithms are less likely to become trapped in local posterior modes than 
 MCMC\, and since it is not necessary to evaluate the full likelihood at ea
 ch step of the algorithm\, a considerable computational saving can also be
  made.\n\nThe class of SMC algorithms of interest in this presentation are
  those that employ MCMC kernels for particle dynamics\, but here\, the ker
 nel moves will be made adaptive. The advantage of adaptive over fixed MCMC
  kernels has been demonstrated in the literature and in particular\, adapt
 ation should be advantageous when the posterior is not Gaussian in shape. 
 This presentation will introduce an SMC method that permits MCMC kernel ch
 oice and adaptive tuning\; the method will be applied to the non-trivial e
 xample of Bayesian mixture analysis.
LOCATION:LR4\, Engineering\, Department of
END:VEVENT
END:VCALENDAR