Fuzzy Classification Approaches Based On AFS Fuzzy Logic

AFS theory was proposed first by professor Liu Xiaodong in 1995. The meaning of AFS is Axiomatic Fuzzy Sets. It can be used to study the law of human thinking and be easily operated by computers. It makes some of the mechanisms of decompose and composition of human conceptions to be understood with mathematical terms. Also it preferably clarified both the essential mechanism that humans use to describe complex fuzzy concepts that depend upon experiences and intuition and the essential mechanism that humans use to choose the corresponding membership function. AFS theory embodies the abstract relationship between human conceptions using topological molecular lattices. It makes the set up of membership functions and fuzzy logic system full of more objectivity, rigor and unity, All of these result come from the original data .So much theories of AFS has already tried to be applied on clustering, pattern recognition and diagnosis ,etc.In this paper, the AFS Fuzzy logic clustering algorithm (X.D. Liu, W.Wang and T.Y.Chai.IEEE Transaction on Systems, Man, Cybernetics, 2005) have been studied further by the improvement of the algorithm and the application of the algorithm to Iris data (reference ftp://ftp.ics.uci.edu/pub/machine-learning-databases/Iris/). In stead of example of less than 10 samples, we apply the improved algorithm to Iris data which has 150 samples and just the order relationship of the samples on the attributes is used. This study shows that the AFS Fuzzy logic clustering algorithm can obtain much higher reclassification accuracy according to the order relationship. Thus the algorithm can be applied to the data sets in which the attributes are only described by order relationship. And improve and study the fuzzy clustering index defined by X.D.Liu, W.Wang and...