BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Modern Neural Networks: the Hinton Camp - Richard Turner\, Mark va
 n der Wilk
DTSTART:20121115T143000Z
DTEND:20121115T160000Z
UID:TALK41576@talks.cam.ac.uk
CONTACT:Konstantina Palla
DESCRIPTION:Historically\, neural networks with multiple hidden layers hav
 e been avoided because they are difficult to train. For example\, training
  the \nnetworks with back-propagation yielded disappointing results\, whic
 h were often worse than those obtained using shallower models. In this RCC
  \nwe will present an alternative approach to training neural networks - a
  form of deep learning developed by Geoffrey Hinton. The main idea is \nto
  train deep generative models layer-by-layer in a greedy unsupervised fash
 ion making use of a theoretical connection with Restrictive Boltzmann \nMa
 chines (RBM's). This network can then be augmented with an additional laye
 r to perform classification\, which gives a greatly increased performance 
 \nover older training methods. We will focus on showing the need for deep 
 models\, describing practical algorithms\, and unpacking some of the theor
 etical \nanalogies used to justify the design choices.
LOCATION:Engineering Department\, CBL Room 438
END:VEVENT
END:VCALENDAR