A new validity index for fuzzy-possibilistic c-means clustering

In some complicated datasets, due to the existence of noisy data points andoutliers, cluster validity indices can yield conicting results in terms of determining theoptimal number of clusters. this paper presents a new validity index for fuzzy-possibilisticc-means clustering called fuzzy-possibilistic (fp) index, which works well in the presenceof clusters that vary in shape and density. moreover, like most of the clustering algorithms,fuzzy-possibilistic c-means (fpcm) is susceptible to some initial parameters. in thisregard, in addition to the number of clusters, fpcm requires a priori selection of thedegree of fuzziness (m) and the degree of typicality (). therefore, an ecient procedurewas presented for determining optimal values of m and . the proposed approach isevaluated using several synthetic and real-world datasets. final computational resultsdemonstrate the capabilities and reliability of the proposed approach compared with severalwell-known fuzzy validity indices in the literature. furthermore, to clarify the ability of theproposed method in real applications, the proposed method is implemented in microarraygene expression data clustering and medical image segmentation.