基于模糊概念相似性与模糊熵度量的模糊分类算法

在AFS(axiomatic fuzzy set)理论框架下,提出了一种基于模糊概念相似性与模糊熵度量的分类算法.模糊分类规则的前件通过概念聚合得到,一种基于模糊概念相似性与模糊熵度量的概念选择函数指导聚合过程;然后,利用剪枝算法对得到的模糊规则集进行剪枝,得到最终的分类规则集.用8组来自UCI数据库的数据集作为实验数据对算法进行验证,并与7种经典分类方法进行比较.实验结果表明该算法能得到较高的分类精度,分类结果明显优于参照的分类方法.

Fuzzy classification algorithm based on fuzzy concept similarity and fuzzy entropy measure

A method to construct a fuzzy concept similarity and fuzzy entropy measure-based classifier by using the axiomatic fuzzy set (AFS) theory is developed. A selection index based on fuzzy concept similarity and fuzzy entropy measure is proposed. Being guided by the selection index, the antecedents of the fuzzy classification rules are selected from the fuzzy concepts which are found when using the aggregation algorithm. And then, the obtained fuzzy rules are pruned by pruning algorithm, and the final classification rule group is obtained. The performance of the proposed classifier is compared with the results produced by 7 classifiers commonly encountered in the literatures when using eight datasets taken from the UCI Machine Learning Repository. It has been found that the accuracy on test data produced by the proposed classifier is higher than that produced by the other classifiers.