蚁群K_medoids融合的聚类算法.pdfVIP

第 26 卷 第 9 期 电子测量与仪器学报 Vol. 26 No. 9 · 800 · JOURNAL OF ELECTRONIC MEASUREMENT AND INSTRUMENT 2012 年 9 月 DOI: 10.3724/SP.J.1187.2012.00800 蚁群K-medoids 融合的聚类算法* 1 2 赵 烨 黄泽君 (1. 合肥工业大学计算机与信息学院, 合肥 230009; 2. 上海大学通信与信息工程学院, 上海 200072) 摘 要: 蚁群算法能够在没有任何先验知识和人为干预的情况下实现自主聚类, 并且鲁棒性较强, 易于与其他算法相结 合。但蚁群算法消耗时间成本较大, 效率较低。而 K-medoids 聚类是一个基于划分的经典聚类算法, 该算法聚类速度快、聚类 效果好而被广泛应用于各种聚类处理中。但需要人为确定簇数目, 并对初始簇中心的依赖性较强。针对以上问题, 提出了结合 蚁群算法和 K-medoids 的聚类算法(AKCA), 该算法融合了蚁群算法和 K-medoids 算法各自在聚类上的优点。实验结果表明,该 算法对于小型数据集具有运行效率高、聚类质量好和自适用性强等优点。 关键词: 聚类分析; 蚁群算法; K-medoids 算法 中图分类号: TP391 文献标识码: A 国家标准学科分类代码: 510.4050 Clustering algorithm based on fusion of ant colony algorithm and K-medoids Zhao Ye1 Huang Zejun2 (1. School of Computer and Information, Hefei University of Technology, Hefei 230009, China; 2. School of Communication and Information Engineering, Shanghai University, Shanghai 200072, China) Abstract: Ant colony algorithm can achieve autonomous clustering without any prior knowledge and human intervention. It is strong robust and easy to combine with other algorithms. But ant colony algorithm is expensive on time consuming. K-medoids algorithm is a classical clustering algorithm based on partitioning. It is widely used because it has high speed and good efficiency. But the number of clusters must be prior decided. K-medoids algorithm dependents on the initial cluster centre points. In order to resolve these problems, a clustering algorithm named ant colony algorithm and K-medoids clustering algorithm (AKCA) is proposed. The advantage of ant col