鲁棒的非负监督低秩鉴别嵌入算法

doi:10.11959/j.issn.2096-6652.202135

Abstract

Abstract:

Non-negative matrix factorization (NMF) has been widely used.However, NMF pays more attention to the local information of the data, it ignores the global representation of the data.In terms of image classification, the global information of the data is often more robust to noise than the local information.In order to improve the robustness of the algorithm, combined with the data of local and global representation, and considered the characteristics of low-rank representation, a non-negative supervised low-rank discriminant embedded algorithm was proposed.This algorithm assumed the existence of noise in the data, decomposed the data into clean data and noise data, and made sparse constraints on the noise matrix through the L₁norm, so as to enhance the robustness to noise.In addition, the algorithm used low-rank representation to learn a low-rank matrix, then through non-negative decomposition, the robustness of the algorithm was enhanced again.Finally, combined with a study of graph embedding method, the local and global data were retained at the same time.The algorithm is applied to various noisy databases, and the recognition rate of this algorithm is improved by about 5%～15% compared with the comparison algorithm.

Key words: non-negative matrix factorization, low-rank representation, feature extraction, graph embedding

CLC Number:

TP391.4

Yu YAO,Minghua WAN. Robust non-negative supervised low-rank discriminant embedding algorithm[J]. Chinese Journal of Intelligent Science and Technology, 2021, 3(3): 342-350.

Figures/Tables 9

References 18

[1]	任克强, 张静然 . 基于局部Fisher准则判别投影的人脸识别算法[J]. 传感器与微系统, 2019,38(7): 113-116,120.
	REN K Q , ZHANG J R . Face recognition algorithm based on local Fisher criterion discriminant projections[J]. Transducer and Microsystem Technologies, 2019,38(7): 113-116,120.
[2]	YANG J , ZHANG D , YANG J Y . Constructing PCA baseline algorithms to reevaluate ICA-based face-recognition performance[J]. IEEE Transactions on Systems,Man,and Cybernetics,Part B (Cybernetics), 2007,37(4): 1015-1021.
[3]	刘文华, 李浥东, 王涛 ,等. 基于高维特征表示的交通场景识别[J]. 智能科学与技术学报, 2019,1(4): 392-399.
	LIU W H , LI Y D , WANG T ,et al. Transportation scene recognition based on high level feature representation[J]. Chinese Journal of Intelligent Science and Technology, 2019,1(4): 392-399.
[4]	ZUO W M , ZHANG D , YANG J ,et al. BDPCA plus LDA:a novel fast feature extraction technique for face recognition[J]. IEEE Transactions on Systems,Man,and Cybernetics,Part B (Cybernetics), 2006,36(4): 946-953.
[5]	NIE F P , WANG Z , WANG R ,et al. Adaptive local linear discriminant analysis[J]. ACM Transactions on Knowledge Discovery from Data, 2020,14(1): 1-19.
[6]	HOYER P O , . Non-negative sparse coding[C]// Proceedings of the 12th IEEE Workshop on Neural Networks for Signal Processing. Piscataway:IEEE Press, 2002: 557-565.
[7]	CAI D , HE X F , HAN J W ,et al. Graph regularized nonnegative matrix factorization for data representation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011,33(8): 1548-1560.
[8]	GUAN N Y , TAO D C , LUO Z G ,et al. Manifold regularized discriminative nonnegative matrix factorization with fast gradient descent[J]. IEEE Transactions on Image Processing, 2011,20(7): 2030-2048.
[9]	杜海顺, 张旭东 . 图嵌入正则化投影非负矩阵分解人脸图像特征提取[J]. 中国图象图形学报, 2014,19(8): 1176-1184.
	DU H S , ZHANG X D . Graph embedding regularized projective non-negative matrix factorization for face image feature extraction[J]. Journal of Image and Graphics, 2014,19(8): 1176-1184.
[10]	ZHANG L J , CHEN Z G , ZHENG M ,et al. Robust non-negative matrix factorization[J]. Frontiers of Electrical and Electronic Engineering in China, 2011,6(2): 192-200.
[11]	KONG D G , DING C , HUANG H . Robust nonnegative matrix factorization using L_2.1-norm[C]// Proceedings of the 20th ACM International Conference on Information and Knowledge Management. New York:ACM Press, 2011.
[12]	WRIGHT J , MA Y , MAIRAL J ,et al. Sparse representation for computer vision and pattern recognition[J]. Proceedings of the IEEE, 2010,98(6): 1031-1044.
[13]	WRIGHT J , YANG A Y , GANESH A ,et al. Robust face recognition via sparse representation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2009,31(2): 210-227.
[14]	YANG M , ZHANG L . Gabor feature based sparse representation for face recognition with Gabor occlusion dictionary[C]// Proceedings of the European Conference on Computer Vision.[S.l.:s.n.], 2010.
[15]	LIU G C , LIN Z C , YAN S C ,et al. Robust recovery of subspace structures by low-rank representation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2013,35(1): 171-184.
[16]	LIU G C , LIN Z C , YONG Y . Robust subspace segmentation by low-rank representation[C]// Proceedings of the International Conference on Machine Learning.[S.l.:s.n.], 2003: 1615-1618.
[17]	LIU G C , LIN Z C , YONG Y . Robust subspace segmentation by low-rank representation[C]// Proceedings of the ICML.[S.l.:s.n.], 2010: 663-670.
[18]	MAZUMDER R , HASTIE T , TIBSHIRANI R . Spectral regularization algorithms for learning large incomplete matrices[J]. The Journal of Machine Learning Research, 2010,11: 2287-2322.

Metrics

Recommended 0

No Suggested Reading articles found!

人脸数据库	干净数据	遮挡块为5×5像素	遮挡块为10×10像素
ORL
Yale
AR
FERET

算法	块大小为5×5				块大小为10×10
算法	t=3	t=4	t=5	t=6	t=3	t=4	t=5	t=6
LDA	57.87%	64.37%	70.83%	76.51%	56.42%	60.21%	66.83%	70.21%
LRR	68.21%	73.33%	75.67%	81.25%	67.21%	72.36%	76.53%	80.67%
LPP	66.71%	72.34%	74.5%	80.21%	64.27%	69.55%	72.61%	78.25%
SLPP	73.66%	77.11%	78.11%	83.06%	67.34%	72.6%	78.07%	81.95%
NMF	70.08%	75.99%	78.91%	85.05%	65.5%	70.17%	73.62%	83.25%
RNMF	78.81%	81.66%	83.71%	85.55%	73.46%	78.21%	83.21%	86.53%
LRNF	82.01%	84.1%	86.25%	89.66%	75.67%	82%	85.37%	88.41%
NSLRE	86.07%	88.33%	90.61%	93.58%	83.05%	86.58%	88.51%	92.12%

算法	遮挡块大小为5×5				遮挡块大小为10×10
算法	t=3	t=4	t=5	t=6	t=3	t=4	t=5	t=6
LDA	62.11%	66.1%	72.33%	76.11%	60.7%	64.1%	70.11%	73.99%
LRR	72.16%	75.23%	78.66%	82.22%	71.42%	73.33%	76.16%	81.11%
LPP	65.65%	70.45%	75.11%	79.31%	64.1%	69%	74.3%	77.8%
SLPP	71.21%	73.66%	77.55%	84.21%	72.06%	75.75%	78.31%	85.65%
NMF	69.06%	72.08%	76.08%	82.25%	68.26%	71.88%	77.33%	80.25%
RNMF	83.33%	85.71%	88.2%	90.5%	81.66%	82.71%	84.88%	89%
LRNF	86.12%	88%	91%	92.12%	80%	84.71%	86.2%	90.21%
NSLRE	90.16%	92.28%	94.3%	95.6%	84.71%	87.3%	90%	92.12%

算法	遮挡块大小为5×5				遮挡块大小为10×10
算法	t=3	t=4	t=5	t=6	t=3	t=4	t=5	t=6
LDA	42.14%	50.72%	53%	58.5%	39.97%	46.45%	50.61%	54%
LRR	61.68%	65.11%	69.85%	74.93%	58.75%	62.11%	66.25%	71.99%
LPP	58.17%	63.90%	67.9%	72%	55.21%	60.06%	63.45%	69%
SLPP	64.05%	67.91%	71.2%	75.98%	61.26%	65.77%	69.21%	73.11%
NMF	62.33%	68.11%	72.25%	77.25%	61.07%	66.21%	71.33%	75.81%
RNMF	68.21%	72.11%	76.25%	79%	65.33%	69.25%	73.67%	76%
LRNF	70.15%	74%	77.15%	82.97%	70%	75.11%	76.87%	80.46%
NSLRE	75.23%	78.16%	82.33%	85.5%	72.17%	77.04%	80.95%	83.25%

算法	遮挡块大小为5×5				遮挡块大小为10×10
算法	t=2	t=3	t=4	t=5	t=2	t=3	t=4	t=5
LDA	41%	42%	46.75%	48.3%	38.55%	41.14%	44.5%	47.33%
LRR	59.21%	61.36%	66.21%	71.17%	56.09%	58.11%	65.34%	69.09%
LPP	57.8%	59.5%	64.25%	69%	54.1%	56.5%	63%	67%
SLPP	57.4%	64.2%	67.6%	73.9%	55.8%	62.03%	68.77%	72.95%
NMF	59.05%	65.01%	71.98%	74.09%	57.11%	62.61%	65.30%	72.34%
RNMF	65.39%	70.25%	77.65%	80.5%	63.21%	66.96%	74.11%	78.10%
LRNF	70.2%	73.1%	80.62%	86.21%	68.36%	71%	76.33%	82.25%
NSLRE	75.23%	78.16%	82.33%	87.60%	73.32%	76.61%	81.45%	86.12%

Robust non-negative supervised low-rank discriminant embedding algorithm

RichHTML

PDF下载

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 9

References 18

Related Articles 10

Metrics

Recommended 0

算法	原始图像	遮挡块大小为5×5	遮挡块大小为10×10
LDA	73.51%	68.21%	62.52%
LRR	77.11%	73.32%	69.15%
LPP	81.24%	76.56%	73.34%
SLPP	80.27%	77.15%	71.69%
NMF	85.63%	83.12%	76.87%
RNMF	89.92%	85.61%	79.13%
LRNF	90.45%	85.1%	82.92%
NSLRE	93.31%	88.70%	84.30%

[1]	Hang YU, Yanwei FU, Boyan JIANG, Xiangyang XUE. A survey of image-based few-shot 3D reconstruction [J]. Chinese Journal of Intelligent Science and Technology, 2022, 4(4): 544-559.
[2]	Hong SHAO, Mingkun ZHANG, Wencheng CUI. Classification method of dermoscopic image based on hierarchical convolution neural network [J]. Chinese Journal of Intelligent Science and Technology, 2021, 3(4): 474-481.
[3]	Heyang WANG, Qiming YANG, Qi ZHU. Retinal multi-disease screening and recognition method based on deep convolution ensemble network [J]. Chinese Journal of Intelligent Science and Technology, 2021, 3(3): 259-267.
[4]	Ying LI, Long CHEN, Zhaohong HUANG, Yang SUN, Guorong CAI. Plant leaf detection technology based on multi-scale CNN feature fusion [J]. Chinese Journal of Intelligent Science and Technology, 2021, 3(3): 304-311.
[5]	Qing TIAN, Rong HU, Zuoyong LI, Yuanzheng CAI, Zhaochai YU. Insulator detection based on SE-YOLOv5s [J]. Chinese Journal of Intelligent Science and Technology, 2021, 3(3): 312-321.
[6]	Qinru YU, Guifu LU. Collaborative representation based classifier with maximum correntropy criterion and locality constraint [J]. Chinese Journal of Intelligent Science and Technology, 2021, 3(3): 334-341.
[7]	Linrui SHI, Yijing HUANG, Jinwu FU, Xinyue GUO, Zizhu FAN. Sparse representation for image recognition based on semi-genetic algorithm in feature space [J]. Chinese Journal of Intelligent Science and Technology, 2021, 3(3): 359-369.
[8]	Wencheng CUI, Pengxia ZHANG, Hong SHAO. Dermoscopic image lesion segmentation method based on deep separable convolutional network [J]. Chinese Journal of Intelligent Science and Technology, 2020, 2(4): 385-393.
[9]	Hongqing DU,Dewang CHEN,Yunhu HUANG,Fenghua ZHU,Lingxi LI. A fuzzy system optimization modeling method based on improved genetic algorithm and support degree [J]. Chinese Journal of Intelligent Science and Technology, 2020, 2(2): 179-185.
[10]	Dewang CHEN,Jijie CAI,Yunhu HUANG. Development prospect of fuzzy system oriented to interpretable artificial intelligence and big data [J]. Chinese Journal of Intelligent Science and Technology, 2019, 1(4): 327-334.