A new look at discriminative training for hidden Markov models

被引：9

作者：

He, Xiaodong ^{[1
]}

Deng, Li ^{[1
]}

机构：

[1] Microsoft Res, Redmond, WA 98052 USA

来源：

PATTERN RECOGNITION LETTERS | 2007年 / 28卷 / 11期

关键词：

hidden Markov model; discriminative learning; minimum classification error; extended Baum-Welch algorithm; growth transformation;

D O I：

10.1016/j.patrec.2006.11.022

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

Discriminative training for hidden Markov models (HMMs) has been a central theme in speech recognition research for many years. One most popular technique is minimum classification error (NICE) training, with the objective function closely related to the empirical error rate and with the optimization method based traditionally on gradient descent. In this paper, we provide a new look at the MCE technique in two ways. First, we develop a non-trivial framework in which the MCE objective function is re-formulated as a rational function for multiple sentence-level training tokens. Second, using this novel re-formulation, we develop a new optimization method for discriminatively estimating HMM parameters based on growth transformation or extended Baum-Welch algorithm. Technical details are given for the use of lattices as a rich representation of competing candidates for the MCE training. (c) 2007 Elsevier B.V. All rights reserved.

引用

页码：1285 / 1294

页数：10

共 50 条

[1] A discriminative training algorithm for hidden Markov models
Ben-Yishai, A
Burshtein, D
[J]. IEEE TRANSACTIONS ON SPEECH AND AUDIO PROCESSING, 2004, 12 (03): : 204 - 217
[2] DISCRIMINATIVE TRAINING FOR BAYESIAN SENSING HIDDEN MARKOV MODELS
Saon, George
Chien, Jen-Tzung
[J]. 2011 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING, 2011, : 5316 - 5319
[3] Discriminative training of hidden Markov models for multiple pitch tracking
Bach, FR
Jordan, MI
[J]. 2005 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING, VOLS 1-5: SPEECH PROCESSING, 2005, : 489 - 492
[4] COMPETITIVE TRAINING - A CONNECTIONIST APPROACH TO THE DISCRIMINATIVE TRAINING OF HIDDEN MARKOV-MODELS
YOUNG, SJ
[J]. IEE PROCEEDINGS-I COMMUNICATIONS SPEECH AND VISION, 1991, 138 (01): : 61 - 68
[5] Large scale discriminative training of hidden Markov models for speech recognition
Woodland, PC
Povey, D
[J]. COMPUTER SPEECH AND LANGUAGE, 2002, 16 (01): : 25 - 47
[6] Discriminative training of hidden Markov models using a classification measure criterion
Chesta, C
Girardi, A
Laface, P
Nigra, M
[J]. PROCEEDINGS OF THE 1998 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING, VOLS 1-6, 1998, : 449 - 452
[7] Discriminative training methods for hidden Markov models: Theory and experiments with perceptron algorithms
Collins, M
[J]. PROCEEDINGS OF THE 2002 CONFERENCE ON EMPIRICAL METHODS IN NATURAL LANGUAGE PROCESSING, 2002, : 1 - 8
[8] Discriminative training of hidden Markov models by multiobjective optimization for visual speech recognition
Lee, JS
Park, CH
[J]. PROCEEDINGS OF THE INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN), VOLS 1-5, 2005, : 2053 - 2058
[9] Large-margin discriminative training of hidden Markov models for speech recognition
Yu, Dong
Deng, Li
[J]. ICSC 2007: INTERNATIONAL CONFERENCE ON SEMANTIC COMPUTING, PROCEEDINGS, 2007, : 429 - +
[10] Markov models - training and evaluation of hidden Markov models
Grewal, Jasleen K.
Krzywinski, Martin
Altman, Naomi
[J]. NATURE METHODS, 2020, 17 (02) : 121 - 122

← 1 2 3 4 5 →