Approximate Low-Rank Projection Learning for Feature Extraction

被引:100
作者
Fang, Xiaozhao [1 ]
Han, Na [1 ]
Wu, Jigang [1 ]
Xu, Yong [2 ,3 ]
Yang, Jian [4 ]
Wong, Wai Keung [5 ,6 ]
Li, Xuelong [7 ,8 ]
机构
[1] Guangdong Univ Technol, Sch Comp Sci & Technol, Guangzhou 510006, Guangdong, Peoples R China
[2] Harbin Inst Technol, Shenzhen Grad Sch, Biocomp Res Ctr, Shenzhen 518055, Peoples R China
[3] Key Lab Network Oriented Intelligent Computat, Shenzhen 518055, Peoples R China
[4] Nanjing Univ Sci & Technol, Sch Comp Sci & Technol, Nanjing 210094, Jiangsu, Peoples R China
[5] Hong Kong Polytech Univ, Inst Text & Clothing, Hong Kong, Hong Kong, Peoples R China
[6] Hong Kong Polytech Univ, Shenzhen Res Inst, Shenzhen 518055, Peoples R China
[7] Chinese Acad Sci, Xian Inst Opt & Precis Mech, Xian 710119, Shaanxi, Peoples R China
[8] Univ Chinese Acad Sci, Beijing 100049, Peoples R China
基金
中国国家自然科学基金;
关键词
Computer vision; feature extraction; low-rank representation (LRR); pattern recognition; ridge regression; FACE RECOGNITION; K-SVD; ILLUMINATION; DICTIONARY; REPRESENTATION; SCENE; GRAPH;
D O I
10.1109/TNNLS.2018.2796133
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Feature extraction plays a significant role in pattern recognition. Recently, many representation-based feature extraction methods have been proposed and achieved successes in many applications. As an excellent unsupervised feature extraction method, latent low-rank representation (LatLRR) has shown its power in extracting salient features. However, LatLRR has the following three disadvantages: 1) the dimension of features obtained using LatLRR cannot be reduced, which is not preferred in feature extraction; 2) two low-rank matrices are separately learned so that the overall optimality may not be guaranteed; and 3) LatLRR is an unsupervised method, which by far has not been extended to the supervised scenario. To this end, in this paper, we first propose to use two different matrices to approximate the low-rank projection in LatLRR so that the dimension of obtained features can be reduced, which is more flexible than original LatLRR. Then, we treat the two low-rank matrices in LatLRR as a whole in the process of learning. In this way, they can be boosted mutually so that the obtained projection can extract more discriminative features. Finally, we extend LatLRR to the supervised scenario by integrating feature extraction with the ridge regression. Thus, the process of feature extraction is closely related to the classification so that the extracted features are discriminative. Extensive experiments are conducted on different databases for unsupervised and supervised feature extraction, and very encouraging results are achieved in comparison with many state-of-the-arts methods.
引用
收藏
页码:5228 / 5241
页数:14
相关论文
共 57 条
[1]   K-SVD: An algorithm for designing overcomplete dictionaries for sparse representation [J].
Aharon, Michal ;
Elad, Michael ;
Bruckstein, Alfred .
IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2006, 54 (11) :4311-4322
[2]  
[Anonymous], 2006, P IEEE C COMPUTER VI, DOI DOI 10.1109/CVPR.2006.301
[3]  
[Anonymous], 1999, Athena scientific Belmont
[4]  
[Anonymous], 24 CVC
[5]  
[Anonymous], 2008, 2008 IEEE Conference on Computer Vision and Pattern Recognition, DOI DOI 10.1109/CVPR.2008.4587408
[6]  
[Anonymous], 2008, P IEEE C COMP VIS PA, DOI DOI 10.1109/CVPR.2008.4587841
[7]  
[Anonymous], 2009, P BRIT MACH VIS C
[8]   In defense of Nearest-Neighbor based image classification [J].
Boiman, Oren ;
Shechtman, Eli ;
Irani, Michal .
2008 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, VOLS 1-12, 2008, :1992-+
[9]   Learning Mid-Level Features For Recognition [J].
Boureau, Y-Lan ;
Bach, Francis ;
LeCun, Yann ;
Ponce, Jean .
2010 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2010, :2559-2566
[10]   Graph Regularized Nonnegative Matrix Factorization for Data Representation [J].
Cai, Deng ;
He, Xiaofei ;
Han, Jiawei ;
Huang, Thomas S. .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2011, 33 (08) :1548-1560