BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Warped Mixture Models for Meaningful Clustering and Bayesian Manif
 old Learning - David Duvenaud\, University of Cambridge
DTSTART:20130311T140000Z
DTEND:20130311T150000Z
UID:TALK43687@talks.cam.ac.uk
CONTACT:Microsoft Research Cambridge Talks Admins
DESCRIPTION:A mixture of Gaussians fit to a single curved or heavy-tailed 
 cluster will report that the data contains many clusters.  To produce mean
 ingful clusterings\, we introduce a model which warps a latent mixture of 
 Gaussians to produce nonparametric cluster shapes.  The possibly low-dimen
 sional latent mixture model allows us to summarize the properties of the h
 igh-dimensional nonlinear clusters (or manifolds)\, whose number\, shape a
 nd dimension is inferred automatically.  We also discuss the pros and cons
  of Hamiltonian Monte Carlo versus variational inference in this nonparame
 tric Bayesian model.
LOCATION:Auditorium\, Microsoft Research Ltd\, 21 Station Road\, Cambridge
 \, CB1 2FB
END:VEVENT
END:VCALENDAR