A fuzzy clustering algorithm using a multiparametric point-to-set distance

We propose a fuzzy classification algorithm that yields a c-class partition P = {A(1),A(2,)...,A(c)} of the data set X = {x(1).x(2),...x(N)}subset ofR(k) Unlike the Fuzzy-Isodata algorithm, (or FCM) where each cluster A(i) has a prototype described only by the mean parameter L-i, the proposed model uses the fuzzy-mean Li, the fuzzy-covariance matrix C-i, and the fuzzy-cardinality n(i). A new formula is used for computing the distance of a point Xi E X to a cluster Ai E P and also for describing file similarity function. The paper presents some experimental results, which prove the effectiveness of the proposed algorithm by comparison to some classical algorithms.