A comparison of training approaches for discriminative segmental models

Segmental models such as segmental conditional random fields have had some recent success in lattice rescoring for speech recognition. They provide a flexible framework for incorporating a wide range of features across different levels of units, such as phones and words. However, such models have mainly been trained by maximizing conditional likelihood, which may not be the best proxy for the task loss of speech recognition. In addition, there has been little work on designing cost functions as surrogates for the word error rate. In this paper, we investigate various losses and introduce a new cost function for training segmental models. We compare lattice rescoring results for multiple tasks and also study the impact of several choices required when optimizing these losses.

[1]  津田 宏治,et al.  International Conference on Machine Learning (ICML)-2005 , 2006 .

[2]  Geoffrey Zweig,et al.  A segmental CRF approach to large vocabulary continuous speech recognition , 2009, 2009 IEEE Workshop on Automatic Speech Recognition & Understanding.

[3]  Lawrence K. Saul,et al.  Large Margin Hidden Markov Models for Automatic Speech Recognition , 2006, NIPS.

[4]  Hui Jiang,et al.  Large margin hidden Markov models for speech recognition , 2006, IEEE Transactions on Audio, Speech, and Language Processing.

[5]  Zdravko Kacic,et al.  A novel loss function for the overall risk criterion based discriminative training of HMM models , 2000, INTERSPEECH.

[6]  Mark J. F. Gales,et al.  Structured Support Vector Machines for Noise Robust Continuous Speech Recognition , 2011, INTERSPEECH.

[7]  Martin A. Riedmiller,et al.  A direct adaptive method for faster backpropagation learning: the RPROP algorithm , 1993, IEEE International Conference on Neural Networks.

[8]  Georg Heigold,et al.  Modified MMI/MPE: a direct evaluation of the margin in speech recognition , 2008, ICML '08.

[9]  Speech Recognition with Segmental Conditional Random Fields , 2016 .

[10]  Mark J. F. Gales,et al.  Structured Log Linear Models for Noise Robust Speech Recognition , 2010, IEEE Signal Processing Letters.

[11]  Dong Yu,et al.  Large-Margin Minimum Classification Error Training for Large-Scale Speech Recognition Tasks , 2007, 2007 IEEE International Conference on Acoustics, Speech and Signal Processing - ICASSP '07.

[12]  Alexander J. Smola,et al.  Tighter Bounds for Structured Estimation , 2008, NIPS.

[13]  Geoffrey Zweig,et al.  Classification and recognition with direct segment models , 2012, 2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[14]  Yoram Singer,et al.  Adaptive Subgradient Methods for Online Learning and Stochastic Optimization , 2011, J. Mach. Learn. Res..

[15]  Sophia Ananiadou,et al.  Stochastic Gradient Descent Training for L1-regularized Log-linear Models with Cumulative Penalty , 2009, ACL.

[16]  Noah A. Smith,et al.  Structured Ramp Loss Minimization for Machine Translation , 2012, HLT-NAACL.

[17]  Gregory Shakhnarovich,et al.  Fingerspelling Recognition with Semi-Markov Conditional Random Fields , 2013, 2013 IEEE International Conference on Computer Vision.

[18]  Geoffrey E. Hinton,et al.  Speech recognition with deep recurrent neural networks , 2013, 2013 IEEE International Conference on Acoustics, Speech and Signal Processing.

[19]  James R. Glass.  A probabilistic framework for segment-based speech recognition , 2003, Comput. Speech Lang..

[20]  Geoffrey Zweig,et al.  SCARF: a segmental conditional random field toolkit for speech recognition , 2010, INTERSPEECH.

[21]  Karen Livescu,et al.  Discriminative Pronunciation Modeling: A Large-Margin, Feature-Rich Approach , 2012, ACL.

[22]  Mark J. F. Gales,et al.  Derivative kernels for noise robust ASR , 2011, 2011 IEEE Workshop on Automatic Speech Recognition & Understanding.

[23]  Georg Heigold,et al.  Discriminative Training for Automatic Speech Recognition: Modeling, Criteria, Optimization, Implementation, and Performance , 2012, IEEE Signal Processing Magazine.

[24]  Ben Taskar,et al.  Learning structured prediction models: a large margin approach , 2005, ICML.

[25]  Thomas Hofmann,et al.  Large Margin Methods for Structured and Interdependent Output Variables , 2005, J. Mach. Learn. Res..

[26]  Mari Ostendorf,et al.  A stochastic segment model for phoneme-based continuous speech recognition , 1989, IEEE Trans. Acoust. Speech Signal Process..

[27]  William W. Cohen,et al.  Semi-Markov Conditional Random Fields for Information Extraction , 2004, NIPS.

[28]  William J. Byrne,et al.  Lattice segmentation and minimum Bayes risk discriminative training for large vocabulary continuous speech recognition , 2006, Speech Commun..

[29]  Geoffrey Zweig,et al.  Integrating meta-information into exemplar-based speech recognition with segmental conditional random fields , 2011, 2011 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[30]  Lalit R. Bahl,et al.  Maximum mutual information estimation of hidden Markov model parameters for speech recognition , 1986, ICASSP '86. IEEE International Conference on Acoustics, Speech, and Signal Processing.

[31]  Lin Xiao,et al.  Dual Averaging Methods for Regularized Stochastic Learning and Online Optimization , 2009, J. Mach. Learn. Res..

[32]  Daniel Povey,et al.  Minimum Phone Error and I-smoothing for improved discriminative training , 2002, 2002 IEEE International Conference on Acoustics, Speech, and Signal Processing.

[33]  Brian Kingsbury,et al.  Boosted MMI for model and feature-space discriminative training , 2008, 2008 IEEE International Conference on Acoustics, Speech and Signal Processing.

[34]  Patrick Wambacq,et al.  Template-Based Continuous Speech Recognition , 2007, IEEE Transactions on Audio, Speech, and Language Processing.

[35]  Andrew McCallum,et al.  Conditional Random Fields: Probabilistic Models for Segmenting and Labeling Sequence Data , 2001, ICML.

[36]  Ambuj Tewari,et al.  Stochastic methods for l1 regularized loss minimization , 2009, ICML '09.