Symmetry as a new measure for cluster validity
Chien Hsing Chou, Mu Su, Eugene Lai · 2002
Abstract: In this paper, a cluster validity measure is presented to infer the appropriateness of data partitions. The proposed validity measure adopts a novel non-metric distance measure based on the idea of "point symmetry". The proposed validity measure can be applied in finding the number of clusters of different geometrical structures. The performance evaluation of the validity measure compares favorably to that of several validity functions and shows the effectiveness. Key-words: cluster validity, clustering algorithm, pattern recognition, similarity measure 1.