BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Parameter estimation in deep learning architectures: Two new insig
 hts. - Prof. Nando de Freitas (Oxford)
DTSTART:20140228T120000Z
DTEND:20140228T130000Z
UID:TALK51164@talks.cam.ac.uk
CONTACT:Zoubin Ghahramani
DESCRIPTION:The talk has two new insights. First\, I will show that it is 
 possible to train most deep learning approaches -regardless of the choice 
 of regularization\, architecture\, algorithms and datasets - by learning o
 nly a small number of the parameters and predicting the rest with nonparam
 etric methods. Often\, this approach makes it possible to learn only 10% o
 f the parameters without a drop in accuracy. Second\, I will introduce a n
 ew method (LAP) for parameter estimation in loopy undirected probabilistic
  graphical models\, also known as Markov random fields\, that is linear in
  the number of cliques\, embarrassingly parallel\, data efficient\, and st
 atistically efficient.
LOCATION: Cambridge University Engineering Department\, LR6
END:VEVENT
END:VCALENDAR