Inductive transfer learning for unlabeled target-domain via hybrid regularization

doi:10.1007/s11434-009-0171-x

	Inductive transfer learning for unlabeled target-domain via hybrid regularization
	Zhuang FuZhen 1,3; Luo Ping 2; He Qing 1; Shi ZhongZhi 1
	2009-07-01
发表期刊	CHINESE SCIENCE BULLETIN
ISSN	1001-6538
卷号	54 期号:14 页码:2470-2478
摘要	Recent years have witnessed an increasing interest in transfer learning. This paper deals with the classification problem that the target-domain with a different distribution from the source-domain is totally unlabeled, and aims to build an inductive model for unseen data. Firstly, we analyze the problem of class ratio drift in the previous work of transductive transfer learning, and propose to use a normalization method to move towards the desired class ratio. Furthermore, we develop a hybrid regularization framework for inductive transfer learning. It considers three factors, including the distribution geometry of the target-domain by manifold regularization, the entropy value of prediction probability by entropy regularization, and the class prior by expectation regularization. This framework is used to adapt the inductive model learnt from the source-domain to the target-domain. Finally, the experiments on the real-world text data show the effectiveness of our inductive method of transfer learning. Meanwhile, it can handle unseen test points.
关键词	Transfer Learning Inductive Learning Transductive Learning Hybrid Regularization
DOI	10.1007/s11434-009-0171-x
收录类别	SCI
语种	英语
WOS研究方向	Science & Technology - Other Topics
WOS类目	Multidisciplinary Sciences
WOS记录号	WOS:000268572300020
出版者	SCIENCE PRESS
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/2XEOYT63/11578
专题	中国科学院计算技术研究所期刊论文_英文
通讯作者	Zhuang FuZhen
作者单位	1.Chinese Acad Sci, Key Lab Intelligent Informat Proc, Inst Comp Technol, Beijing 100190, Peoples R China 2.Hewlett Packard Labs China, Beijing 100084, Peoples R China 3.Chinese Acad Sci, Grad Univ, Beijing 100190, Peoples R China
推荐引用方式 GB/T 7714	Zhuang FuZhen,Luo Ping,He Qing,et al. Inductive transfer learning for unlabeled target-domain via hybrid regularization[J]. CHINESE SCIENCE BULLETIN,2009,54(14):2470-2478.
APA	Zhuang FuZhen,Luo Ping,He Qing,&Shi ZhongZhi.(2009).Inductive transfer learning for unlabeled target-domain via hybrid regularization.CHINESE SCIENCE BULLETIN,54(14),2470-2478.
MLA	Zhuang FuZhen,et al."Inductive transfer learning for unlabeled target-domain via hybrid regularization".CHINESE SCIENCE BULLETIN 54.14(2009):2470-2478.