WSEAS Transactions on Signal Processing
Print ISSN: 1790-5052, E-ISSN: 2224-3488Volume 11, 2015
Learning One-class KSVM+ for Multi-class Problems with Group Information of Data
Authors: ,
Abstract: This paper is denoted to study the effect of the group information of data in one-class kernel support vector machines (OC-KSVMs) for classification accuracy and time consumed of multi-class classification data. Two new classification methods based on OC-KSVMs are presented. One is OC-KSVM with maximum margin from the origin and group information of data (briefly, MMOC-KSVM+) and another is OC-KSVM with hypersphere and group information of data (briefly, HSOC-KSVM+). We proved theoretically that MMOC-KSVM and HSOC-KSVM are equivalent for Gaussian RBF kernels. Experiments on three real-words data sets are performed in order to test and evaluate the efficacy of the proposed methods. Experimental results indicate that the group information of data can improve the classification accuracy of data and meanwhile increase the time consumed of algorithms.
Keywords: multi-class classification problem, One-class kernel SVM, group information of data, maximum margin, hyper-sphere
Pages: 262-271
WSEAS Transactions on Signal Processing, ISSN / E-ISSN: 1790-5052 / 2224-3488, Volume 11, 2015, Art. #31