BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Large Margin Training of Hidden Markov Models - Anton Ragni
DTSTART:20090219T140000Z
DTEND:20090219T153000Z
UID:TALK15408@talks.cam.ac.uk
CONTACT:Shakir Mohamed
DESCRIPTION:Large Margin Training of Hidden Markov Models in Speech Recogn
 ition.\n\nIn particular\, I will make first an introduction to automatic s
 peech recognition followed by some popular approaches to train such system
 s (ML\, MMIE). Next I will highlight the main weaknesses of them and intro
 duce some of the alternative training frameworks. Mainly I will focus on L
 arge Margin Training applied to HMMs. Finally\, I will give a description 
 of the state of the art system used to transcribe broadcast news in three 
 languages (English\, Arabic and Mandarin) developed here at University of 
 Cambridge.\n\nThe paper on the Large Margin Training:\n\nFei Sha\, Lawrenc
 e K. Saul\, "Large Margin Hidden Markov Models for Automatic Speech Recogn
 ition"\, NIPS 2006\, http://books.nips.cc/papers/files/nips19/NIPS2006_014
 3.pdf\n\nThe corresponding PhD thesis containing sequential derivation of 
 Large Margin Training algorithms for GMM\, observable Markov Model and\, f
 inally\, Hidden Markov Model:\n\nFei Sha\, "Large Margin Training of Acous
 tic Models for Speech Recognition"\, University of Pennsylvania\, 2007. \n
 http://www-rcf.usc.edu/~feisha/pubs/thesis_tree.pdf\n\nShort summary on La
 rge Margin training of Gaussian Mixture Models:\n\nFei Sha\, Lawrence K. S
 aul\, " Large Margin Gaussian Mixture Modeling for Phonetic Classification
  and Recognition"\, Proc. ICASSP\, 2006. \nhttp://ieeexplore.ieee.org/stam
 p/stamp.jsp?arnumber=1660008&isnumber=34757\n
LOCATION:Engineering Department\, CBL Room 438
END:VEVENT
END:VCALENDAR