الفهرس | Only 14 pages are availabe for public view |
Abstract Clustering is one of the main research directions in data mining (DM) based on unsupervised learning. It aims at partitioning a given set of objects into homogenous groups such that the objects of the same cluster share similar characteristics (high intra-cluster similarity or cohesiveness within clusters), relative to those belonging to different clusters (low inter-cluster similarity or distinctiveness between clusters). The concept of min-plus algebra ( * + ), or ( ), involves combining two operators ―min‖ and ―plus‖ on a basic set . The minimum operator chooses the smaller one from two elements, * + The plus operator adds two elements, . It is most natural to associate Euclidean distance metric with the min-plus metric in order to take the minimum of all distances between a data point towards all cluster centroids. The thrust of the proposed work is to determine which measure is suitable for data clustering amongst 12 distance/similarity measures. |