BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Deep Learning of Natural Language Semantics - Prof Yoshua Bengio -
  Université de Montréal
DTSTART:20150413T130000Z
DTEND:20150413T140000Z
UID:TALK58643@talks.cam.ac.uk
CONTACT:David Greaves
DESCRIPTION:The core ingredient of deep learning is the notion of distribu
 ted representation.  This talk will start by explaining its theoretical ad
 vantages for modelling language\, in comparison with non-parametric method
 s based on counting frequencies of occurrence of observed tuples of values
  (like with n-grams). The talk will then explain how having multiple level
 s of representation\, i.e.\, depth\, can in principle give another exponen
 tial advantage. Neural language models have been extremely successful in r
 ecent years but extending their reach from language modelling to machine t
 ranslation is very appealing because it forces the learned intermediate re
 presentations to capture meaning\, and we found that the resulting word em
 beddings are qualitatively different. Recently\, we introduced the notion 
 of attention-based neural machine translation\, with impressive results on
  several language pairs\, and we applied the same ideas to caption generat
 ion\, i.e.\, generating a sentence associated with an input image\, yieldi
 ng impressive results that will conclude the talk.\n\nYoshua Bengio is\, a
 long with Geoff Hinton and Yann LeCun\, one of the machine learning resear
 chers who has led the recent deep learning renaissance. He has made signif
 icant contributions to the understanding of deep neural networks and his w
 ork has led to advances in numerous application areas including computer v
 ision\, artificial intelligence and natural language understanding. He has
  won numerous awards for his work and his research has been cited over 22\
 ,000 times. 
LOCATION:Lecture Theatre 1\, Computer Laboratory
END:VEVENT
END:VCALENDAR