An Improved Support Vector Machine Using Class-Median Vectors

摘要：

Abstract Support vector machine builds the final discriminant function on only a small part of the training samples, which may make the decision rule too sensitive to noise and outliers. Inspired by the idea of central support vector machine or CSVM, we present an improved method based on the class-median, called Median Support Vector Machine or MSVM in this paper. The experiment results show MSVM is a promising and robust algorithm, especially when outliers are far from the class-center.

关键词： Pattern Classification Support Vector Machine Median Support Vector Machine

作者: Zhenzhen Kou Jianhua Xu Xuegong Zhang Liang Ji

作者单位: State Key Laboratory of Intelligent Technology and Systems Department of Automation, Tsinghua University, Beijing 100084, P.R.C.

会议类型: 国际会议

会议名称: 8th International Conference on Neural Information Processing(ICONIP 2001)(第八届国际神经信息处理大会)

会议地点: 上海

会议语种:英文

页码: 921-925

在线出版日期: 2001-11-14（万方平台首次上网日期，不代表论文的发表时间）