In order to solve the problem of multiple classification,based on the idea of explicit and implicit factor in factor space theory,the algorithm of sweeping class serial classification is proposed,the merging of categories is defined,and an explicit and implicit method of merging class sweep classification is proposed.The algorithm steps are given and analyzed by numerical examples.The pair wise combination is defined and the pairwise class sweep classification method with explicit and implicit factors is proposed,the algorithm steps are given and analyzed by numerical examples.During the execution of two algorithms,for the occurrence of decision categories can not be separated,the absolute value method of difference with explicit and implicit factors is proposed to solve the problem.Three examples in UCI dataset are compared with support vector machine,the results show that the proposed method of merging class sweep classification and the pairwise sweep classification method can realize factor explicit and implicit,and the accuracy of classification algorithm is better than that of the support vector machine.The conclusion of explicit and implicit study of factors in multi-classification learning expands the theoretical and applied research of factor space.
关键词
因素空间/因素显隐/扫类连环分类算法/合并扫类连环分类算法/两两扫类连环分类算法/差额绝对值法
Key words
factor space/factor explicit implicit/class sweep algorithm/merging class sweep classification algorithm/pairwise class sweep classification algorithm/difference absolute value method
维普
万方数据