Learning policy scheduling for text augmentation

doi:10.1016/j.neunet.2021.09.028

	Learning policy scheduling for text augmentation
	Li, Shuokai ; Ao, Xiang 1; Pan, Feiyang ; He, Qing
	2022
发表期刊	NEURAL NETWORKS
ISSN	0893-6080
卷号	145 页码:121-127
摘要	When training deep learning models, data augmentation is an important technique to improve the performance and alleviate overfitting. In natural language processing (NLP), existing augmentation methods often use fixed strategies. However, it might be preferred to use different augmentation policies in different stage of training, and different datasets may require different augmentation policies. In this paper, we take dynamic policy scheduling into consideration. We design a search space over augmentation policies by integrating several common augmentation operations. Then, we adopt a population based training method to search the best augmentation schedule. We conduct extensive experiments on five text classification and two machine translation tasks. The results show that the optimized dynamic augmentation schedules achieve significant improvements against previous methods. (C) 2021 Elsevier Ltd. All rights reserved.
关键词	Data augmentation Text classification
DOI	10.1016/j.neunet.2021.09.028
收录类别	SCI
语种	英语
资助项目	National Key Research and De-velopment Program of China[2017YFB1002104] ; National Natural Science Foundation of China[92046003] ; National Natural Science Foundation of China[61976204] ; National Natural Science Foundation of China[U1811461] ; Project of Youth Innovation Promotion Association CAS ; Beijing Nova Program[Z201100006820062] ; Ant Financial through the Ant Financial Science Funds for Security Research
WOS研究方向	Computer Science ; Neurosciences & Neurology
WOS类目	Computer Science, Artificial Intelligence ; Neurosciences
WOS记录号	WOS:000717665500006
出版者	PERGAMON-ELSEVIER SCIENCE LTD
引用统计	被引频次：6[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/18112
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Ao, Xiang
作者单位	1.Chinese Acad Sci, CAS, Inst Comp Technol, Key Lab Intelligent Informat Proc, Beijing, Peoples R China 2.Univ Chinese Acad Sci, Beijing, Peoples R China 3.Inst Intelligent Comp Technol, Suzhou, Peoples R China
推荐引用方式 GB/T 7714	Li, Shuokai,Ao, Xiang,Pan, Feiyang,et al. Learning policy scheduling for text augmentation[J]. NEURAL NETWORKS,2022,145:121-127.
APA	Li, Shuokai,Ao, Xiang,Pan, Feiyang,&He, Qing.(2022).Learning policy scheduling for text augmentation.NEURAL NETWORKS,145,121-127.
MLA	Li, Shuokai,et al."Learning policy scheduling for text augmentation".NEURAL NETWORKS 145(2022):121-127.