Authors: Weijie Chen; Yuanhai Shao; Yibo Jiang; Chongpu Xia
Addresses: Zhijiang College, Zhejiang University of Technology, Hangzhou, Zhejiang, China ' Zhijiang College, Zhejiang University of Technology, Hangzhou, Zhejiang, China ' College of Computer Science and Technology, Zhejiang University of Technology, Hangzhou, Zhejiang, China ' Department of Power Distribution Network, NARI Technology Development Co., Ltd., China
Abstract: In this paper, to improve the generalisation ability of generalised eigenvalues proximal support vector machines (GEPSVM), we propose an ensemble GEPSVM, called EnGEP for short. Note that GEPSVM is not sensitive to different weights of the points, to increase the potential diversity of GEPSVM, firstly, we introduce an extra parameter in GEPSVM, which gives different penalties for two non-hyperplanes determines by GEPSVM. Then, we use a novel bagging strategy to ensemble GEPSVM with additional parameters. Experimental results both on artificial and benchmark datasets show that our EnGEP improves the generalisation performance of GEPSVM greatly, and it also reveals the effectiveness of our EnGEP.
Keywords: pattern classification; ensemble learning; GEPSVM; nonparallel hyperplanes; artificial intelligence; generalised eigenvalues; support vector machines; SVM; generalisation ability.
International Journal of Computer Applications in Technology, 2013 Vol.47 No.2/3, pp.273 - 279
Received: 08 May 2021
Accepted: 12 May 2021
Published online: 05 Jun 2013 *