BSC: Belief Shift Clustering

It is still a challenging problem to characterize uncertainty and imprecision between specific (singleton) clusters with arbitrary shapes sizes. In order solve such problem, we propose belief shift clustering (BSC) method for dealing object data. The BSC considered as the evidential version of mean or mode seeking under theory functions. First, new notion, called shift, provided preliminarily assign each query noise, precise, imprecise one. Second, rule designed partial credal redistribution object. To avoid “uniform effect” useless calculations, dynamic framework simulated cluster centers established reassign singleton related meta-cluster. Once an assigned meta-cluster, this may be in overlapping intermediate areas different clusters. Consequently, can reasonably effectiveness has been verified on several artificial, natural, image segmentation/classification datasets by comparison other methods.