BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Spectral Learning and Decoding for Natural Language Parsing - Shay
  Cohen\, University of Edinburgh
DTSTART:20140124T120000Z
DTEND:20140124T130000Z
UID:TALK49973@talks.cam.ac.uk
CONTACT:Tamara Polajnar
DESCRIPTION:Spectral methods have received considerable attention in the p
 ast in the machine learning and the NLP communities. Most recently\, they\
 nhave been applied to latent-variable modelling. In this case\, their two\
 nmost clear advantages over the use of EM is their computational efficienc
 y and the sound theory behind them (they are not prone to local maxima lik
 e EM).\n\nIn this talk\, I will present two distinct uses of the spectral 
 method for\nnatural language parsing. I will describe a learning algorithm
  for latent-variable PCFGs\, a very useful model for constituent parsing. 
 I will also describe our use of tensor decomposition for speeding up parsi
 ng inference. Here\, we approximate the underlying model by using a tensor
  decomposition algorithm\, and this approximation permits us to use fast i
 nference with dynamic programming.\n\nIf time permits\, I will also touch 
 on the use of spectral decomposition algorithms for unsupervised learning.
 \n\nJoint work with Michael Collins\, Dean Foster\, Ankur Parikh\, Giorgio
  Satta\, Karl Stratos\, Lyle Ungar\, Eric Xing
LOCATION:FW26\, Computer Laboratory
END:VEVENT
END:VCALENDAR