文献类型：期刊文献

英文题名：Nonpeaked Discriminant Analysis for Data Representation

作者：Ye, Qiaolin[1,2] Li, Zechao[3] Fu, Liyong[2,4] Zhang, Zhao[5,6] Yang, Wankou[7] Yang, Guowei[8,9]

第一作者：Ye, Qiaolin

通信作者：Fu, LY[1]

机构：[1]Nanjing Forestry Univ, Coll Informat Sci & Technol, Nanjing 210037, Jiangsu, Peoples R China;[2]Chinese Acad Forestry, Inst Forest Resource Informat Tech, Beijing 100091, Peoples R China;[3]Nanjing Univ Sci & Technol, Coll Comp Sci & Technol, Nanjing 210094, Jiangsu, Peoples R China;[4]Natl Forestry & Grassland Adm, Key Lab Forest Management & Growth Modeling, Beijing 100091, Peoples R China;[5]Soochow Univ, Sch Comp Sci & Technol, Suzhou 215000, Peoples R China;[6]Hefei Univ Technol, Sch Comp & Informat, Hefei 230000, Anhui, Peoples R China;[7]Southeast Univ, Sch Automat, Nanjing 210096, Jiangsu, Peoples R China;[8]Nanjing Audit Univ, Jiangsu Key Lab Auditing Informat Engn, Nanjing 211815, Jiangsu, Peoples R China;[9]Nanjing Audit Univ, Sch Informat Engn, Nanjing 211815, Jiangsu, Peoples R China

年份：2019

卷号：30

期号：12

起止页码：3818-3832

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

收录：;EI(收录号：20195207901289);Scopus(收录号：2-s2.0-85076447067);WOS:【SCI-EXPANDED(收录号:WOS:000502762600025)】；

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 61871444, in part by the National Key Research and Development Program of China under Grant 2017YFC0820601, in part by the Natural Science Foundation of Jiangsu Province under Grant BK20171453 and Grant BK20170033, in part by the National Science Foundations of China under Grant 61672365, Grant 61772277, Grant 61773117, and Grant 61772275, and in part by the Qinglan and Six Talent Peaks Projects of Jiangsu Province. (Qiaolin Ye and Zechao Li contributed equally to this work.)

语种：英文

外文关键词：Cutting L-norm distance; data classification; discriminant analysis; robustness

摘要：Of late, there are many studies on the robust discriminant analysis, which adopt L-1-norm as the distance metric, but their results are not robust enough to gain universal acceptance. To overcome this problem, the authors of this article present a nonpeaked discriminant analysis (NPDA) technique, in which cutting L-1-norm is adopted as the distance metric. As this kind of norm can better eliminate heavy outliers in learning models, the proposed algorithm is expected to be stronger in performing feature extraction tasks for data representation than the existing robust discriminant analysis techniques, which are based on the L-1-norm distance metric. The authors also present a comprehensive analysis to show that cutting L-1-norm distance can be computed equally well, using the difference between two special convex functions. Against this background, an efficient iterative algorithm is designed for the optimization of the proposed objective. Theoretical proofs on the convergence of the algorithm are also presented. Theoretical insights and effectiveness of the proposed method are validated by experimental tests on several real data sets.