一种基于旋转超盒和引力场融合的聚类算法

A Novel Clustering Method Based-on Rotated Super-Box and Gravitation Fusion

针对传统聚类算法处理复杂分布数据的不足,提出了一种新型的基于旋转超盒和引力场融合的聚类算法.该算法由1)数据集归一化;2)利用旋转超盒构造初始类别;3)借助引力场概念对初始类别进行融合3个步骤构成.仿真结果表明,该算法在无需聚类数目的情况下,对复杂分布数据具有很好的聚类效果.

To the shortage of traditional clustering algorithm when dealing data with complicated distribution, a novel hierarchical clustering method based-on the rotated super-box and gravitation fusion(RBGFCA) is presented in this paper. This algorithm consists of three parts: (a)uniform for data; (b) constitution of initial patterns by rotated super-box; (c)fusion of initial patterns with gravitation. The simulation results show that compared to FCA, this algorithm has good clustering performance for data with complicated distribution without knowing the number of clustering.