在阐述二值支持向量机分类器的基础上,针对M类问题重点讨论了一对多与一对一分类器组合求解方案。对于纽合方案中出现的不可分类问题,论述了引入模糊隶属函数、构造有向非循环图等相关的解决思路,针对标准数据库分类问题比较了不同组合方案的分类性能。 Based on the explanation of the binary SVM classifier, a one-to-one and one-to-one classifier combination solution is mainly discussed for the M class problem. For the unclassifiable problems in the cooperation project, this paper discusses the related solutions of introducing fuzzy membership functions and constructing directed acyclic graphs. The classification performance of different combination schemes is compared with the standard database classification problems.