A comparative study of RPCL and MCE based discriminative training methods for LVCSR

Authors:
Zaihu Pang;Xihong Wu;Lei Xu
Affiliations:
Speech and Hearing Research Center, Key Laboratory of Machine Perception, (Ministry of Education), Peking University, China;Speech and Hearing Research Center, Key Laboratory of Machine Perception, (Ministry of Education), Peking University, China;Speech and Hearing Research Center, Key Laboratory of Machine Perception, (Ministry of Education), Peking University, China,Department of Computer Science and Engineering, The Chinese University o ...
Venue:
IScIDE'11 Proceedings of the Second Sino-foreign-interchange conference on Intelligent Science and Intelligent Data Engineering
Year:
2011

Citing 5
Cited 1

Comparison of discriminative training criteria and optimization methods for speech recognition

Speech Communication
A unified perspective and new results on RHT computing, mixture based learning, and multi-learner based problem solving

Pattern Recognition
Discriminative training of HMMs for automatic speech recognition: A survey

Computer Speech and Language
Discriminative learning for minimum error classification [patternrecognition]

IEEE Transactions on Signal Processing
Rival penalized competitive learning for clustering analysis, RBF net, and curve detection

IEEE Transactions on Neural Networks

Discriminative GMM-HMM acoustic model selection using two-level bayesian ying-yang harmony learning

IScIDE'12 Proceedings of the third Sino-foreign-interchange conference on Intelligent Science and Intelligent Data Engineering

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper presents a comparative study of two discriminative methods, i.e., Rival Penalized Competitive Learning (RPCL) and Minimum Classification Error (MCE), for the tasks of Large Vocabulary Continuous Speech Recognition (LVCSR). MCE aims at minimizing a smoothed sentence error on training data, while RPCL focus on avoiding misclassification through enforcing the learning of correct class and de-learning its best rival class. For a fair comparison, both the two discriminative mechanisms are implemented at state level. The LVCSR results show that both MCE and RPCL perform better than Maximum Likelihood Estimation (MLE), while RPCL has better discriminative and generative abilities than MCE.