文献类型：期刊文献

英文题名：Learning Robust Discriminant Subspace Based on Joint L-2,L- p- and L-2,L-s-Norm Distance Metrics

作者：Fu, Liyong[1,2,3] Li, Zechao[4] Ye, Qiaolin[2] Yin, Hang[2] Liu, Qingwang[3,5] Chen, Xiaobo[6] Fan, Xijian[2] Yang, Wankou[7] Yang, Guowei[8]

第一作者：Fu, Liyong;符利勇

通信作者：Ye, QL[1]

机构：[1]Cent South Univ Forestry & Technol, Res Ctr Forestry Remote Sensing & Informat Engn, Changsha 410004, Peoples R China;[2]Nanjing Forestry Univ, Coll Informat Sci & Technol, Nanjing 210037, Peoples R China;[3]Chinese Acad Forestry, Inst Forest Resource Informat Tech, Beijing 100091, Peoples R China;[4]Nanjing Univ Sci & Technol, Coll Comp Sci & Technol, Nanjing 210094, Peoples R China;[5]Natl Forestry & Grassland Adm, Key Lab Forest Management & Growth Modeling, Beijing 100091, Peoples R China;[6]Jiangsu Univ, Automot Engn Res Inst, Zhenjiang 212013, Jiangsu, Peoples R China;[7]Southeast Univ, Sch Automat, Nanjing 210096, Peoples R China;[8]Nanjing Audit Univ, Sch Informat Engn, Jiangsu Key Lab Auditing Informat Engn, Nanjing 211815, Peoples R China

年份：2022

卷号：33

期号：1

起止页码：130-144

外文期刊名：IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS

收录：;WOS:【SCI-EXPANDED(收录号:WOS:000739635300016)】；

基金：This work was supported in part by the Central Public-Interest Scientific Institution Basal Research Fund under Grant CAFYBB2019QD003, in part by the National Science Foundation of China under Grant 62072246, in part by the Natural Science Foundation of Jiangsu Province under Grant BK20171453, in part by the National Science Foundations of China under Grant 61773210 and Grant 61773117, and in part by the Qinglan and Six Talent Peaks Projects of Jiangsu Province.

语种：英文

外文关键词：Discriminant power; joint L-2,L- p- and L-2,L-s-norm; nongreedy; robust discriminant analysis (RDA); rotational invariance; s-norm distance metrics

摘要：Recently, there are many works on discriminant analysis, which promote the robustness of models against outliers by using L-1- or L-2,L-1-norm as the distance metric. However, both of their robustness and discriminant power are limited. In this article, we present a new robust discriminant subspace (RDS) learning method for feature extraction, with an objective function formulated in a different form. To guarantee the subspace to be robust and discriminative, we measure the within-class distances based on L-2,L-s-norm and use L-2,L- p-norm to measure the between-class distances. This also makes our method include rotational invariance. Since the proposed model involves both L-2,(p)-norm maximization and L-2,L-s-norm minimization, it is very challenging to solve. To address this problem, we present an efficient nongreedy iterative algorithm. Besides, motivated by trace ratio criterion, a mechanism of automatically balancing the contributions of different terms in our objective is found. RDS is very flexible, as it can be extended to other existing feature extraction techniques. An in-depth theoretical analysis of the algorithm's convergence is presented in this article. Experiments are conducted on several typical databases for image classification, and the promising results indicate the effectiveness of RDS.