基于特征空间聚类的二叉树支持向量机分类算法

Feature-based Spatial Clustering of the Binary Support Vector Machine Classification Algorithm

采用数据挖掘中聚类分析的类距离定义,在高维特征空间中,计算各类别间的最短距离,以最短距离作为该类与其他类的距离,提出了一种基于聚类-二叉树支持向量机分类算法。该算法能够简化计算,同时通过类距离比较实现了对类距离最大者的优先分离,实验结果表明该算法具有一定的优越性。

Cluster analysis using data mining from the definition of the class,in hingh-dimensional feature space,calculating the shortest distance between the various categories,the shortest distance as the distence between the class and others.Proposed a clustering-based-binary tree support vector machine classification algorithm.The algorithm can simplify the calculation,experimental results show that the algorithm has some advantages.