基于可扩展函数族的聚类算法的改进

在原有基于可扩展函数族聚类的基础上,提出了自适应可扩展函数族概念,对原来的算法CIFF和CDFF作了改进,将阈值理论与可扩展函数族相结合,设计了新的聚类算法,并对其聚类性能作了分析。实验结果表明,用自适用可扩展函数族方法进行聚类,不但使聚类在可伸缩性、增量数据处理及复杂数据类型处理等方面都表现出很好的性能,而且与原算法相比,具有聚类精度高、速度快等优点。

Improvement of Cluster Algorithm Based on the Extensible Function Family

A conception of self-adaptive extensible function family is presented on the basis of original extensible function family-based cluster.It improves the original algorithms CIFF and CDFF.A new cluster algorithm is designed by combining threshold value theory with extensible function family and its cluster performance is analyzed.Experimental results show that the cluster made by self-adaptive extensible function family has not only excellent performances in aspects of extensibility,incremental data processing and complex data type processing,etc.,but also such merits as highly precision and fast velocity in cluster processing comparing with CIFF and CDFF.