BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:An Introduction to Sum Product Networks - José Miguel Hernández 
 Lobato (University of Cambridge)
DTSTART:20130328T150000Z
DTEND:20130328T163000Z
UID:TALK43634@talks.cam.ac.uk
CONTACT:Colorado Reed
DESCRIPTION:Sum product networks (SPNs) are a new family of deep probabili
 stic models in which exact inference is tractable. SPNs are directed acycl
 ic graphs with variables as leaves\, sums and products as internal nodes\,
  and weighted edges. A SPN is an arithmetic circuit which under some condi
 tions (completeness and consistency) represents the partition function and
  all marginals of some graphical model. Essentially all tractable graphica
 l models can be cast as SPNs\, but SPNs are also more general. Discriminat
 ive and generative learning of SPNs can be efficiently implemented using h
 ard EM and hard gradient descent. These methods avoid the problem of gradi
 ent diffusion in deep architectures and allow us to effectively work with 
 SPNs of more than 30 layers of hidden variables. Several experiments show 
 that SPNs have state of the art performance on different image completion 
 and classification tasks\, outperforming alternative deep and shallow meth
 ods.
LOCATION:Engineering Department\, CBL Room 438
END:VEVENT
END:VCALENDAR