期刊
IEEE JOURNAL OF SELECTED TOPICS IN SIGNAL PROCESSING
卷 4, 期 6, 页码 917-925出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/JSTSP.2010.2076110
关键词
Handwriting recognition; large-vocabulary continuous speech recognition; margin-based training; part-of-speech tagging
资金
- OSEO
- French State agency for innovation
- European Union [FP6-033549]
- Defense Advanced Research Projects Agency (DARPA) [HR001-06-C-0023]
Typical training criteria for string recognition like for example minimum phone error (MPE) and maximum mutual information (MMI) in speech recognition are based on a (regularized) loss function. In contrast, large-margin classifiers-the de-facto standard in machine learning-maximize the separation margin. An additional loss term penalizes misclassified samples. This paper shows how typical training criteria like for example MPE or MMI can be extended to incorporate the margin concept, and that such modified training criteria are smooth approximations to support vector machines with the respective loss function. The proposed approach takes advantage of the generalization bounds of large-margin classifiers while keeping the efficient framework for conventional discriminative training. This allows us to directly evaluate the utility of the margin term for string recognition. Experimental results are presented using the proposed modified training criteria for different tasks from speech recognition (including large-vocabulary continuous speech recognition tasks trained on up to 1500-h audio data), part-of-speech tagging, and handwriting recognition.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据