Study Of Fuzzy Association Classification Based On AFS Fuzzy Logic

AFS theory(Axiomatic Fuzzy Sets) is a new method to study fuzzy set which was proposed by professor Liu Xiaodong in 1995. AFS theory have studied and discussed the essential problems of fuzzy theory: How to find the strict and consistent algorithms of determining membership functions for fuzzy concepts and the fuzzy logical operations accurately representing human thinking logic. In the framework of AFS theory, AFS present a new algorithm of determining membership functions for fuzzy concepts according to original date and information and propose AFS fuzzy logic. In fact, AFS fuzzy logic is more appropriate to represent human thinking logic and the models of intelligent systems in real world applications based on original data and information, which are comprehensible and have definitive semantic meanings. These approaches have potential applications in the study of recognition and large-scale complicate intelligent systems. Recently, AFS theory has been developed further and applied to fuzzy clustering analysis, fuzzy decision trees, credit rating analysis, pattern recognition and hitch diagnoses, etc.In this paper, we combine the AFS fuzzy logic classification algorithm and fuzzy association rule to study further the classifiers based on AFS fuzzy logic. Employing the idea of the AFS fuzzy logic classification algorithm and the methods of designing the fuzzy association classifier, we apply the fuzzy association rules to design a new classifier based on AFS fuzzy logic. In addition, a modification of support commonly used in association classifier is made more reasonable. With MATLAB, we implement the algorithm and experiments on 5 UCI(University of California, Irvine) data sets to validate the performance of the algorithm. Finally, we compare the accuracy of the data sets obtained from the classifier proposed with the results present in literatures. The results show that the proposed classifier is practical and useful.