摘要： 针对解决多类模式识别问题的SVM方法进行研究,在比较常用的几种多类SVM分类算法基础上,提出了一种基于核聚类方法的多层次SVM分类树,将核空间中的无监督学习方法和有监督学习方法结合起来,实现了一种结构更加简洁清晰、计算效率更高的多层SVM分类树算法,并在实验中取得了良好的结果。

关键词: 多类模式识别, 支持向量机, 核聚类, 统计学习理论

Abstract: This paper researches the support vector machines for multiclass pattern recognition and compares some common SVMs for multiclass classification problems. And proposes a hierarchical support vector machines based on kernel cluster method combining the unsupervised learning method and supervised learning together. The algorithm is more effective and simple in structure and proved to performance better by experiment.

中图分类号: 

• TP18