Fixed point relational fuzzy clustering

Abstract

The proposed relational fuzzy clustering method called FRFP (fuzzy relational fixed point) is not based on minimizing an objective function, as in traditional methods, but rather on determining a fixed point of a function of the desired membership matrix with the proximity matrix as parameter. The proposed method is compared to other relational clustering methods including NERFCM, Rouben's method and Windhams AP method. A clustering quality index is calculated for doing the comparison. using various proximity matrices as input. Simulations show the method to be very effective and less computationally expensive than other fuzzy relational data clustering methods.