BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Spectral Learning -  Maxim Rabinovich\, Aman Sinha
DTSTART:20131114T150000Z
DTEND:20131114T163000Z
UID:TALK48883@talks.cam.ac.uk
CONTACT:Konstantina Palla
DESCRIPTION:Over the past few years\, "spectral methods" have been applied
  with great success to parameter estimation in latent variable models. For
  this talk\, we will explain how spectral methods work in two models of wi
 de interest: hidden Markov models (HMMs) and latent-variable probabilistic
  context-free grammars (LPCFGs). In the process\, we will see that these m
 ethods rely only on linear and multilinear algebra\, making them highly ef
 ficient. We will then go on to discuss what is perhaps their most signific
 ant edge: consistency guarantees that are conspicuously absent from classi
 cal estimation techniques like EM (though we will see that the story is mo
 re complicated than one might think). Finally\, we will introduce the unif
 ied theoretical perspective on these algorithms that has recently emerged.
LOCATION:Engineering Department\, CBL Room 438
END:VEVENT
END:VCALENDAR