摘要
提出了一种基于支持向量预选取的两类分类方法,在不影响分类性能的前提下,根据支持向量的几何特征,采用向量投影的概率分布法和减法聚类再选取法,在训练样本集中预先选择与分类有关的一组支持向量,代替整个训练样本集,进行支持向量机两类分类.数据实验结果表明,该方法有效地提高了支持向量机分类器的学习效率,实现了较高的分类精确度.
A support vector machine classification method based on pre-extracting is proposed. In this approach, both the probability distribution of vector projection and the subtraetive clustering are adopted to select the threshold of sample boundary vectors. Then samples are pre-extracted, and the training sample is replaced by the pre-extraeted sample with implementing support vector machine classification constructed. The experimental results show that the proposed method has considerably improved the learning efficiency of the support vector machine classifier and yields higher classification accuracy.
出处
《哈尔滨理工大学学报》
CAS
北大核心
2009年第A01期5-7,10,共4页
Journal of Harbin University of Science and Technology
基金
基金项目:国家自然科学基金(60575036)
“973”国家重点基础研究发展规划(2002cb312200-3)
哈尔滨市科技创新人才研究专项资金项目(2007RFXXG023)
哈尔滨理工大学优秀拔尖创新人才培养基金.
关键词
支持向量机
预选取
边界向量
减法聚类
support vector machine
pre-extracting
boundary vector
subtractive clustering