球形支持向量机是一种学习算法,它通过在高维特征空间中,对每一个模式类别构造一个覆盖其所有训练样本的具有最小体积的超球体,来实现对训练样本空间的划分。在此基础上,提出了一种基于新的决策规则的球形支持向量机算法,并在七个UCI数据集上进行了实验,实验结果表明提出的算法可以取得比标准的支持向量机算法更好的分类效果。 Sphere-based SVM is a learning algorithm, which achieves the partition of training sample space by constructing a microsphere with the smallest volume covering all the training samples for each mode category in high-dimensional feature space. Based on this, a new spherical support vector machine (SVM) algorithm based on the new decision rule is proposed and tested on seven UCI datasets. The experimental results show that the proposed algorithm can achieve better performance than the standard SVM algorithm The classification effect.