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.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Similar content being viewed by others
References
Bahl, L., Brown, P., De Souza, P., Mercer, R.: Maximum mutual information estimation of hidden Markov model parameters for speech recognition. In: Proc. of 1986 IEEE Intl. Conf. on ICASSP, pp. 49–52 (1986)
Juang, B.H., Katagiri, S.: Discriminative learning for Minimum Error Classification. IEEE Trans. on Signal Processing 40(12), 3043–3054 (1992)
Povey, D., Woodland, P.C.: Minimum phone error and I-smothing for improved discriminative training. In: Proc. of 2002 IEEE Intl. Conf. on ICASSP, vol. 1, pp. 105–108 (2002)
Jiang, H.: Discriminative training of HMMs for automatic speech recognition: A survey. Computer Speech and Language 24, 589–608 (2010)
Pang, Z.H., Tu, S.K., Su, D., Wu, X.H., Xu, L.: Discriminative training of GMM-HMM acoustic model by RPCL learning. Frontiers of Electrical and Electronic Engineering in China 6(2), 283–290 (2011)
Schlüter, R., Macherey, W., Müller, B., Ney, H.: Comparison of discriminative training criteria and optimization methods for speech recognition. Speech Communication 34(3), 287–310 (2001)
Fu, Q., He, X., Deng, L.: Phone-discriminating minimum classification error(P-MCE) training criteria for phonetic recognition. In: Proceedings of InterSpeech 2007, pp. 2073–2076 (2007)
Macherey, W., Haferkamp, L., Schlüter, R., Ney, H.: Investigations on error minimizing training criteria for discriminative training in acoustic speech recognition. In: Proceedings of EuroSpeech 2005, Lisbon, Portugal, pp. 2133–2136 (2005)
Yan, Z.J., Zhu, B., Hu, Y., Wang, R.H.: Minmum word classification error training of HMMs for automatic speech recognition. In: Proc. of 2008 IEEE Intl. Conf. on ICASSP, pp. 4521–4524 (2008)
Xu, L., Krzyzak, A., Oja, E.: Unsupervised and supervised classifications by rival penalized competitive learning. In: Proc. 11th Intl. Conf. on Pattern Recognition, Netherlands, vol. I, pp. 672–675 (1992)
Xu, L., Krzyzak, A., Oja, E.: Rival penalized competitive learning for clustering analysis, RBF net, and curve detection. IEEE Trans. on Neural Networks 4(4), 636–649 (1993)
Xu, L.: Rival penalized competitive learning. Scholarpedia 2(8), 1810 (2007)
Xu, L.: A unified perspective and new results on RHT computing, mixture based learning, and multi-learner based problem solving. Pattern Recognition 40(8), 2129–2153 (2007)
Xu, L.: Bayesian Ying-Yang system, best harmony learning, and five action circling. FEEE in China 5(3), 281–328 (2010)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2012 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Pang, Z., Wu, X., Xu, L. (2012). A Comparative Study of RPCL and MCE Based Discriminative Training Methods for LVCSR. In: Zhang, Y., Zhou, ZH., Zhang, C., Li, Y. (eds) Intelligent Science and Intelligent Data Engineering. IScIDE 2011. Lecture Notes in Computer Science, vol 7202. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-31919-8_4
Download citation
DOI: https://doi.org/10.1007/978-3-642-31919-8_4
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-31918-1
Online ISBN: 978-3-642-31919-8
eBook Packages: Computer ScienceComputer Science (R0)