摘要
数据描述只使用目标集训练样本获得关于目标集的描述,支持向量数据描述(SVDD)是一种有效的单值分类数据描述算法.根据分类边界线上的支持向量之间距离的大小,利用距离的相似度来对训练集进行约减.实验结果表明,该算法与传统SVDD相比减少了训练时所需的支持向量数目,因而减少了测试时间,同时分类性能也稍有提高.
By employing training samples of target set only, data description is obtained by description of target set, and SVDD （Support Vector Data Description） is proved an efficient data description algorithm of one-class classification. According to the distance between support vectors located on the classification boundary, this paper uses distance similarity to reduce the training samples. Experiments show that, compared with the traditional SVDD, this algorithm reduces the number of support vectors needed for exercise and thereby reduces the testing time; the classification performance of this algorithm is also increased in some degree.
出处
《五邑大学学报（自然科学版）》
CAS
2008年第2期52-56,共5页
Journal of Wuyi University(Natural Science Edition)
关键词
支持向量数据描述
单值分类
距离相似度
support vector data description （SVDD）
one-class classification
distance similarity
