creators_name: Kanevsky, Dimitri
creators_name: Povey, Daniel
creators_name: Ramabhadran, Bhuvana
creators_name: Rish, Irina
creators_name: Sainath, Tara
creators_id: kanevsky@us.ibm.com
creators_id: dpovey@us.ibm.com
creators_id: bhuvana@us.ibm.com
creators_id: rish@us.ibm.com
creators_id: tsainath@MIT.EDU
type: preprint
datestamp: 2008-04-30 18:35:19
lastmod: 2011-03-11 08:57:06
metadata_visibility: show
title: A New Family of Extended Baum-Welch Update Rules 
subjects: comp-sci-speech
full_text_status: public
keywords: Extended Baum Welch, optimization, speech
abstract: In this paper, we consider a generalization of the state-of-art
discriminative method for optimizing the conditional likelihood in
Hidden Markov Models (HMMs), called the Extended Baum-Welch (EBW)
algorithm, that has had significant impact on the speech recognition
community. We propose a generalized form of EBW update rules that
can be associated with a weighted sum of updated and initial models,
and  demonstrate that using novel update rules can significantly
speed up parameter estimation for Gaussian mixtures.
date: 2008-04-27
date_type: completed
refereed: FALSE
referencetext: Cong Liu   Peng Liu   Hui Jiang   Soong, F.   Ren-Hua Wang, Constrained Line Search Optimization for Discriminative Training in Speech Recognition, Proc. ICASSP, 2007.

Dimitri Kanevsky, Extended Baum Transformations for General Functions, Proc. ICASSP.

Dimitri Kanevsky, Extended Baum Transformations for General Functions, II, tech. Rep. RC23645(W0506-120), Human Language technologies, IBM.

Daniel Povey, Discriminative Training for Large Vocabulary Speech Recognition, PhD Thesis March 1, Cambridge, UK.

Daniel Povey, Brian Kingsbury, Evaluation of Proposed Modifications to MPE for Large Scale Discriminative Training, ICASSP'07.

Daniel Povey, Dimitri Kanevsky, Brian Kingsbury, Bhuvana Ramabhadran, George Saon and Karthik Visweswariah, Boosted MMI for model and feature-space discriminative training, ICASSP'08.

Tara N. Sainath, Dimitri Kanevsky, Bhuvana Ramabhadran, Gradient Steepness Metrics Using Extended Baum-Welch Transformations for Universal Pattern Recognition Tasks, Proc. ASRU, 2007

P.C. Woodland and D. Povey, Large Scale Discriminative Training of hidden Markov modeles for Speech Recognition, Computer Speech and Language, pp. 25-47, Vol. 16, No. 1, January 2002.

citation:   Kanevsky, Dr Dimitri and Povey, Dr Daniel and Ramabhadran, Dr Bhuvana and Rish, Dr Irina and Sainath, Dr Tara  (2008) A New Family of Extended Baum-Welch Update Rules.  [Preprint]     
document_url: http://cogprints.org/6037/1/ebw.pdf