Multi-Objective Matrix Normalization for Fine-Grained Visual Recognition

doi:10.1109/TIP.2020.2977457

	Multi-Objective Matrix Normalization for Fine-Grained Visual Recognition
	Min, Shaobo 1; Yao, Hantao2 ; Xie, Hongtao 1; Zha, Zheng-Jun 1; Zhang, Yongdong 1
发表期刊	IEEE TRANSACTIONS ON IMAGE PROCESSING
ISSN	1057-7149
	2020
卷号	29 页码:4996-5009
通讯作者	Xie, Hongtao(htxie@ustc.edu.cn) ; Zhang, Yongdong(zhyd73@ustc.edu.cn)
摘要	Bilinear pooling achieves great success in fine-grained visual recognition (FGVC). Recent methods have shown that the matrix power normalization can stabilize the second-order information in bilinear features, but some problems, e.g., redundant information and over-fitting, remain to be resolved. In this paper, we propose an efficient Multi-Objective Matrix Normalization (MOMN) method that can simultaneously normalize a bilinear representation in terms of square-root, low-rank, and sparsity. These three regularizers can not only stabilize the second-order information, but also compact the bilinear features and promote model generalization. In MOMN, a core challenge is how to jointly optimize three non-smooth regularizers of different convex properties. To this end, MOMN first formulates them into an augmented Lagrange formula with approximated regularizer constraints. Then, auxiliary variables are introduced to relax different constraints, which allow each regularizer to be solved alternately. Finally, several updating strategies based on gradient descent are designed to obtain consistent convergence and efficient implementation. Consequently, MOMN is implemented with only matrix multiplication, which is well-compatible with GPU acceleration, and the normalized bilinear features are stabilized and discriminative. Experiments on five public benchmarks for FGVC demonstrate that the proposed MOMN is superior to existing normalization-based methods in terms of both accuracy and efficiency. The code is available: https://github.com/mboboGO/MOMN.
关键词	Visualization Graphics processing units Feature extraction Convergence Optimization Covariance matrices Training Fine-grained visual recognition bilinear pooling matrix normalization multi-objective optimization
DOI	10.1109/TIP.2020.2977457
收录类别	SCI
语种	英语
资助项目	National Key Research and Development Program of China[2017YFC0820600] ; National Nature Science Foundation of China[61525206] ; National Nature Science Foundation of China[U1936210] ; National Postdoctoral Programme for Innovative Talents[BX20180358] ; Youth Innovation Promotion Association Chinese Academy of Sciences[2017209] ; Fundamental Research Funds for the Central Universities[WK2100100030]
项目资助者	National Key Research and Development Program of China ; National Nature Science Foundation of China ; National Postdoctoral Programme for Innovative Talents ; Youth Innovation Promotion Association Chinese Academy of Sciences ; Fundamental Research Funds for the Central Universities
WOS研究方向	Computer Science ; Engineering
WOS类目	Computer Science, Artificial Intelligence ; Engineering, Electrical & Electronic
WOS记录号	WOS:000522226700005
出版者	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
引用统计	被引频次：49[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://ir.ia.ac.cn/handle/173211/38714
专题	多模态人工智能系统全国重点实验室_多媒体计算
通讯作者	Xie, Hongtao; Zhang, Yongdong
作者单位	1.Univ Sci & Technol China, Sch Informat Sci & Technol, Hefei 230026, Peoples R China 2.Chinese Acad Sci, Natl Lab Pattern Recognit, Inst Automat, Beijing 100864, Peoples R China
推荐引用方式 GB/T 7714	Min, Shaobo,Yao, Hantao,Xie, Hongtao,et al. Multi-Objective Matrix Normalization for Fine-Grained Visual Recognition[J]. IEEE TRANSACTIONS ON IMAGE PROCESSING,2020,29:4996-5009.
APA	Min, Shaobo,Yao, Hantao,Xie, Hongtao,Zha, Zheng-Jun,&Zhang, Yongdong.(2020).Multi-Objective Matrix Normalization for Fine-Grained Visual Recognition.IEEE TRANSACTIONS ON IMAGE PROCESSING,29,4996-5009.
MLA	Min, Shaobo,et al."Multi-Objective Matrix Normalization for Fine-Grained Visual Recognition".IEEE TRANSACTIONS ON IMAGE PROCESSING 29(2020):4996-5009.