A Novel Incremental SVM Learning Algorithm

摘要：

In this paper we present a novel approach toincremental support vector machine (SVM) learningalgorithm. We analysis the possible change of supportvector set after new samples are added to training set.Based on the analysis result, a novel algorithm ispresented. In this algorithm useless samples arediscarded and knowledge is accumulated. The experimentresult shows that this algorithm is more effective thantraditional SVM while the classification precision is alsoguaranteed.

作者: Zeng Wenhua Ma Jian

作者单位: Departmeny of Computer Science,Xiamen University,P.R.China,361005 School of Computer Science,Hangzhou Institute of Electronics Engineering,HangZhou,P.R.China 310037

会议类型: 国际会议

会议名称: The 8th International Conference on Computer Supported Cooperative Work in Design(第八届计算机支持的协同工作设计国际会议)(CSCWD2004)

会议地点: 厦门

会议语种:英文

页码: 658-662

在线出版日期: 2004-05-26（万方平台首次上网日期，不代表论文的发表时间）