摘要
为了解决方言辨识系统中训练样本冗余的问题,提出了一种融合多样性测度的汉语方言主动辨识方法。利用SVM分类器选取不确定性的样本。根据样本间分布情况的测度算法,选取出兼具多样性的训练样本,经过多次迭代将这些最具区别性的样本组成训练集。将此训练集重新输入到SVM进行分类辨识。实验结果表明,该方法能有效克服选取样本的冗余,与传统的主动学习方法相比,在同等识别率的情况下,人工标注样本的数量减少了50%。
In order to solve the problem of the redundant training samples in dialect identification system,an approach forChinese dialect identification fusing diversity measure is proposed.Firstly,the uncertain samples are chosen by SVMclassifier,then according to the distribution of these samples,the uncertain samples with diversity are selected and thenew training set including these distinctive samples is constructed after several iterations.Finally,SVM is reused to makethe decision.Experimental results indicate that,compared with the traditional active method,the proposed approacheffectively overcomes the redundancy of the samples and the number of manually annotated samples reduces50%underthe same condition of recognition accuracy.
作者
夏玉果
戴红霞
顾明亮
XIA Yuguo;DAI Hongxia;GU Mingliang(School of Electronic Information Engineering, Jiangsu Vocational College of Information Technology, Wuxi, Jiangsu 214153, China;School of Linguistic Science, Jiangsu Normal University, Xuzhou, Jiangsu 221116, China)
出处
《计算机工程与应用》
CSCD
北大核心
2017年第15期149-154,共6页
Computer Engineering and Applications
基金
国家自然科学基金(No.61040053)
关键词
汉语方言辨识
主动学习
支持矢量机
多样性测度
Chinese dialect identification
active learning
Support Vector Machine(SVM)
diversity measure
