An improved SVM algorithm is proposed based on the new concept.With the concepts of class-radius, class-centroid-distance and class-centripetal We can delete those non-SV effectively with high accuracy and generality when the data was promiscuous.The experiments show that our method achieved a satisfactory result.%提出一种改进的支持向量机分类方法.通过引入分类圆心、分类半径、分类圆心距等概念,从而更加快速准确地删除非支持向量点,引入混淆度的概念,解决如何在样本严重混淆时进行剔除混淆点,保证算法的泛化性.实验证明,采用这种改进的算法能够在严重混淆的训练样本中保证准确度的同时提高支持向量机分类速度.
展开▼