面向复杂簇的聚类算法研究与实现

RESEARCH AND IMPLEMENTATION OF COMPLEX CLUSTER ORIENTED CLUSTERING ALGORITHM

有效聚类各种复杂的数据对象簇是聚类算法应用于事务对象划分、图像分割、机器学习等方面需要解决的关键技术。在分析与研究现有聚类算法的基础上,提出一种基于密度和自适应密度可达的改进算法。实验证明,该算法能够有效聚类任意分布形状、不同密度、不同尺度的簇;同时,算法的计算复杂度与传统基于密度的聚类算法相比有明显的降低。

For transaction item classification,image segmentation and machine learning, the key technique is to handle complicatedly distributed clusters efficiently. On the basis of the analysis and research of traditional clustering algorithms, a clustering algorithm based on density and adaptive density-reachable is presented. Experimental results show that the algorithm can handle clusters of arbitrary shapes, sizes and densities. At the same time, compared with other density-based algorithms, this algorithm can evidently reduce computing complexity.