一种新的分裂层次聚类SVM多值分类器  

A Novel SVM Multi-class Classifier Based on Divisive Hierarchical Clustering

作　　者：张国云[1] 章兢[1] 

机构地区：[1]湖南大学电气与信息工程学院

出　　处：《控制与决策》2005年第8期931-934,共4页Control and Decision

基　　金：教育部科学技术研究重点项目([2001]224)

摘　　要：提出一种分裂层次聚类SVM分类树分类方法.该方法通过融合模糊聚类技术和支持向量机算法,利用分裂的层次聚类策略,有选择地重新构造学习样本集和SVM子分类器,得到了一种树形多值分类器.研究结果表明,对于k类别模式识别问题,该方法只需构造k-1个SVM子分类器,克服了SVM子分类器过多以及存在不可区分区域的缺点,具有良好的分类性能.实验结果验证了该方法的优越性.A novel support vector machines classification tree approach based on divisive hierarchical clustering （DHCSVC） is proposed. By fusion of fuzzy clustering techniques and support vector machines （SVM）, and utilizing the divisive hierarchical clustering strategy, this method selectively re-constructs learning samples set and corresponding SVM sub-classifier. A tree-shape multi-class classifier is also built using this method. In k-class task,the new classifier only contains k-1 SVM sub-classifiers. The proposed method also overcomes the drawbacks such as unclassifiable region which the “one-against-one” method has, and possesses a good classification performance.The experimental results show that the proposed approach has superiority in classification task.

关 键 词：分裂层次聚类 支持向量机 多值分类器 分类树 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]