Robust Recovery of Corrupted Low-Rank Matrix by Implicit Regularizers

CORC > 自动化研究所 > 中国科学院自动化研究所 > 智能感知与计算研究中心

	Robust Recovery of Corrupted Low-Rank Matrix by Implicit Regularizers
	He, Ran1,2 ; Tan, Tieniu1,2 ; Wang, Liang1,2
刊名	IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
	2014-04-01
卷号	36 期号:4 页码:770-783
关键词	PCA implicit regularizers low-rank matrix recovery correntropy l(1) regularization
英文摘要	Low-rank matrix recovery algorithms aim to recover a corrupted low-rank matrix with sparse errors. However, corrupted errors may not be sparse in real-world problems and the relationship between l(1) regularizer on noise and robust M-estimators is still unknown. This paper proposes a general robust framework for low-rank matrix recovery via implicit regularizers of robust M-estimators, which are derived from convex conjugacy and can be used to model arbitrarily corrupted errors. Based on the additive form of half-quadratic optimization, proximity operators of implicit regularizers are developed such that both low-rank structure and corrupted errors can be alternately recovered. In particular, the dual relationship between the absolute function in l(1) regularizer and Huber M-estimator is studied, which establishes a connection between robust low-rank matrix recovery methods and M-estimators based robust principal component analysis methods. Extensive experiments on synthetic and real-world data sets corroborate our claims and verify the robustness of the proposed framework.
WOS标题词	Science & Technology ; Technology
类目[WOS]	Computer Science, Artificial Intelligence ; Engineering, Electrical & Electronic
研究领域[WOS]	Computer Science ; Engineering
关键词[WOS]	LINEAR INVERSE PROBLEMS ; THRESHOLDING ALGORITHM ; IMAGE-RESTORATION ; FACE RECOGNITION ; COMPLETION ; SIGNAL ; SHRINKAGE ; FRAMEWORK ; LASSO
收录类别	SCI
语种	英语
WOS记录号	WOS:000334109000011
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/3798]
专题	自动化研究所_智能感知与计算研究中心
作者单位	1.Chinese Acad Sci, Ctr Res Intelligent Percept & Comp CRIPAC, Beijing 100190, Peoples R China 2.Chinese Acad Sci, Natl Lab Pattern Recognit NLPR, Inst Automat, Beijing 100190, Peoples R China
推荐引用方式 GB/T 7714	He, Ran,Tan, Tieniu,Wang, Liang. Robust Recovery of Corrupted Low-Rank Matrix by Implicit Regularizers[J]. IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE,2014,36(4):770-783.
APA	He, Ran,Tan, Tieniu,&Wang, Liang.(2014).Robust Recovery of Corrupted Low-Rank Matrix by Implicit Regularizers.IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE,36(4),770-783.
MLA	He, Ran,et al."Robust Recovery of Corrupted Low-Rank Matrix by Implicit Regularizers".IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE 36.4(2014):770-783.