文献类型：期刊文献

英文题名：An efficient nonlinear one-class SVM based on matrix patterns

作者：Chen, Yanyan[1]; Zhang, Yun[1]; Ling, Ling[1]

第一作者：陈艳燕

通讯作者：Chen, Yanyan

机构：[1] College of Applied Science and Technology, Beijing Union University, No. 97, Beisihuan East Road, Chaoyang District, Beijing, 100101, China

第一机构：北京联合大学应用科技学院

年份：2016

卷号：7

期号：8

起止页码：1657-1664

外文期刊名：ICIC Express Letters, Part B: Applications

收录：EI(收录号：20163302710958);Scopus(收录号：2-s2.0-84981173500)

语种：英文

外文关键词：Support vector machines

摘要：In many fields, we often encounter one-class classification problems. The traditional vector-based one-class classification algorithms represented by one-class SVM (OCSVM) have limitations when matrix is considered as input data. This work addresses a nonlinear one-class classifier with matrix-based maximal margin classification paradigm. To this end, we formulate the Nonlinear One-class SVM Based on Matrix Patterns (NLMatOCSVM), which can directly use matrix as input. That helps to retain the data topology more efficiently in comparison with vector-based classifier. The efficiency of the proposed method is illustrated on two matrix-based human faces datasets. The experimental results indicate the validity of the new method. ? 2016 ICIC International.