Minimum-risk training for semi-Markov conditional random fields with application to handwritten Chinese/Japanese text recognition

ISCAS OpenIR

	Minimum-risk training for semi-Markov conditional random fields with application to handwritten Chinese/Japanese text recognition
	Zhou, Xiang-Dong (1); Zhang, Yan-Ming (2); Tian, Feng (3); Wang, Hong-An (3); Liu, Cheng-Lin (2); Zhou, X.-D.(zhouxiangdong@cigit.ac.cn)
	2014
发表期刊	Pattern Recognition
ISSN	313203
卷号	47 期号:5 页码:1904-1916
摘要	Semi-Markov conditional random fields (semi-CRFs) are usually trained with maximum a posteriori (MAP) criterion which adopts the 0/1 cost for measuring the loss of misclassification. In this paper, based on our previous work on handwritten Chinese/Japanese text recognition (HCTR) using semi-CRFs, we propose an alternative parameter learning method by minimizing the risk on the training set, which has unequal misclassification costs depending on the hypothesis and the ground-truth. Based on this framework, three non-uniform cost functions are compared with the conventional 0/1 cost, and training data selection is incorporated to reduce the computational complexity. In experiments of online handwriting recognition on databases CASIA-OLHWDB and TUAT Kondate, we compared the performances of the proposed method with several widely used learning criteria, including conditional log-likelihood (CLL), softmax-margin (SMM), minimum classification error (MCE), large-margin MCE (LM-MCE) and max-margin (MM). On the test set (online handwritten texts) of ICDAR 2011 Chinese handwriting recognition competition, the proposed method outperforms the best system in competition. © 2013 Elsevier Ltd. All rights reserved.; Semi-Markov conditional random fields (semi-CRFs) are usually trained with maximum a posteriori (MAP) criterion which adopts the 0/1 cost for measuring the loss of misclassification. In this paper, based on our previous work on handwritten Chinese/Japanese text recognition (HCTR) using semi-CRFs, we propose an alternative parameter learning method by minimizing the risk on the training set, which has unequal misclassification costs depending on the hypothesis and the ground-truth. Based on this framework, three non-uniform cost functions are compared with the conventional 0/1 cost, and training data selection is incorporated to reduce the computational complexity. In experiments of online handwriting recognition on databases CASIA-OLHWDB and TUAT Kondate, we compared the performances of the proposed method with several widely used learning criteria, including conditional log-likelihood (CLL), softmax-margin (SMM), minimum classification error (MCE), large-margin MCE (LM-MCE) and max-margin (MM). On the test set (online handwritten texts) of ICDAR 2011 Chinese handwriting recognition competition, the proposed method outperforms the best system in competition. © 2013 Elsevier Ltd. All rights reserved.
收录类别	SCI ; EI
关键词	Semi-markov Conditional Random Fields Minimum-risk Training Character String Recognition
部门归属	(1) Intelligent Media Technique Research Center, Chongqing Institute of Green and Intelligent Technology, Chinese Academy of Sciences, Chongqing 400714, China; (2) National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences, 95 Zhongguan East Road, Beijing 100190, China; (3) Beijing Key Lab of Human-Computer Interaction, Institute of Software, Chinese Academy of Sciences, Beijing 100190, China
语种	英语
WOS记录号	WOS:000331667400011
引用统计
内容类型	期刊论文
URI标识	http://ir.iscas.ac.cn/handle/311060/16721
专题	中国科学院软件研究所
通讯作者	Zhou, X.-D.(zhouxiangdong@cigit.ac.cn)
推荐引用方式 GB/T 7714	Zhou, Xiang-Dong ,Zhang, Yan-Ming ,Tian, Feng ,et al. Minimum-risk training for semi-Markov conditional random fields with application to handwritten Chinese/Japanese text recognition[J]. Pattern Recognition,2014,47(5):1904-1916.
APA	Zhou, Xiang-Dong ,Zhang, Yan-Ming ,Tian, Feng ,Wang, Hong-An ,Liu, Cheng-Lin ,&Zhou, X.-D..(2014).Minimum-risk training for semi-Markov conditional random fields with application to handwritten Chinese/Japanese text recognition.Pattern Recognition,47(5),1904-1916.
MLA	Zhou, Xiang-Dong ,et al."Minimum-risk training for semi-Markov conditional random fields with application to handwritten Chinese/Japanese text recognition".Pattern Recognition 47.5(2014):1904-1916.