CABGSI: An efficient clustering algorithm based on structural information of graphs

Le résumé fourni par la source

This paper introduces CABGSI, a novel graph-based clustering algorithm that effectively addresses the limitations of traditional clustering techniques. Unlike conventional methods that require predefined cluster quantities and assume simple geometrical data structures, CABGSI leverages graph structural entropy to naturally discern clusters. Our algorithm constructs a network among data points, capturing the intrinsic structure of the data. We demonstrate its superior performance on diverse datasets, particularly those with complex geometries such as speckles and images. Despite challenges with moon and ring configurations, CABGSI significantly outperforms traditional algorithms like k-means by not relying on predetermined cluster centers. This innovation broadens the applicability of clustering techniques and paves the way for future research in data analysis.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.