Non-exclusive Clustering: A Partitioning Approach

Abstract

Non-exclusive clustering is a partitioning based clustering scheme wherein the data points are clustered such that they belong to one or more clusters. Usually in real world applications, the datasets that we work with are not entirely exclusive in nature. In applications such as gene expression data analysis and satellite image processing, non-exclusive algorithms need to be employed for better and more accurate cluster analysis. Therefore, we intend to tackle such problems with a non-exclusive clustering algorithm, closely determined by a nonexclusivity score (NES). The NES is based on a feature class correlation measure, which helps to determine the significant overlap between the data points in the dataset and aids us in comprehending the clusters to which they belong to.