VMCFDGS: Variable Multicenter Aggregation Clustering Method Based on Fuzzy Dominating (Dominated)-Granularity Structure

With the continuous surge in data, the order information increases while the distinguishability of the data diminishes. To address the decreased efficiency and stability of traditional clustering methods owing to intercluster overlap and noise, this article proposes a novel method—a variable multicenter aggregation clustering algorithm based on fuzzy dominating (dominated)-granularity structure (VMCFDGS). First, the order characteristics of object attributes are utilized to form a fuzzy dominating (dominated)-granularity structure (FDDGS). Then, we employ the characteristics of the FDDGS to illustrate relations among objects from various perspectives. Furthermore, an approach is proposed that uses a multicenter technique for better description of similarities within and across clusters, clustering data from diverse granularities, and mining clustering outcomes at different levels. Finally, fusion clustering of information purification is achieved based on the optimum results of different granularity clustering on each metric. Comparative results with advanced clustering algorithms on the UCI dataset demonstrate this proposed method's superiority in processing complex spatial structure data and its effectiveness and robustness in handling overlapping and noisy data.