计算机应用 ›› 2014, Vol. 34 ›› Issue (1): 204-207.DOI: 10.11772/j.issn.1001-9081.2014.01.0204

• 人工智能 • 上一篇    下一篇

基于K-means的改进人工蜂群聚类算法

曹永春,蔡正琦,邵亚斌   

  1. 西北民族大学 数学与计算机科学学院,兰州 730030
  • 收稿日期:2013-07-26 修回日期:2013-09-08 出版日期:2014-01-01 发布日期:2014-02-14
  • 通讯作者: 曹永春
  • 作者简介:曹永春(1972-),男,甘肃天祝人,副教授,主要研究方向:智能计算、复杂网络;蔡正琦(1974-),男,甘肃天水人,副教授,主要研究方向:智能信息处理;邵亚斌(1974-),男,甘肃天水人,副教授,博士,主要研究方向:不确定性处理的数学。
  • 基金资助:

    国家自然科学基金资助项目;2012年度国家民委科研资金资助项目;中央高校基本科研业务费专项资金资助项目

Improved artificial bee colony clustering algorithm based on K-means

CAO Yongchun,CAI Zhenqi,SHAO Yabin   

  1. School of Mathematics and Computer Science, Northwest University for Nationalities, Lanzhou Gansu 730030, China
  • Received:2013-07-26 Revised:2013-09-08 Online:2014-01-01 Published:2014-02-14
  • Contact: CAO Yongchun

摘要: 针对K-means聚类算法对初始聚类中心敏感和易陷入局部最优解的缺点,提出一种基于K-means的人工蜂群(ABC)聚类算法。将改进的人工蜂群算法和K-means迭代相结合,使算法对初始聚类中心的依赖性和陷入局部最优解的可能性降低,提高了算法的稳定性。通过基于反向学习的初始化策略,增强了初始群体的多样性。利用非线性选择策略,改善了过早收敛问题,提高了搜索效率。通过对邻域搜索范围的动态调整,提高了算法收敛速度,增强了局部寻优能力。实验结果表明,该算法不仅克服了K-means算法稳定性差的缺点,而且具有良好的性能和聚类效果。

关键词: 人工蜂群算法, 聚类分析, K-means, 反向学习, 非线性选择

Abstract: Since the K-means clustering method is sensitive to initial clustering centers and easy to be trapped by local optimum, an Artificial Bee Colony (ABC) clustering algorithm based on K-means was proposed in this paper. This algorithm integrated the improved ABC algorithm with the K-means iteration, which reduced the dependence on the initial clustering centers and the probability to be trapped by local optimum, thus improving the stability of the algorithm. The initialization strategy based on the opposition-based learning improved the diversity of the initial population. The algorithm overcame the problem of premature convergence and improved the efficiency of searching through introducing nonlinear selection strategy. The convergence speed was accelerated and the capability of local optimization was enhanced by dynamically adjusting the neighborhood search range. The experimental results show that the clustering efficiency and performance has been significantly improved, as well as its stability.

Key words: Artificial Bee Colony (ABC) algorithm, clustering analysis, K-means, opposition-based learning, nonlinear selection

中图分类号: