计算机工程与应用 ›› 2010, Vol. 46 ›› Issue (23): 41-43.DOI: 10.3778/j.issn.1002-8331.2010.23.011

• 研究、探讨 • 上一篇    下一篇

聚类问题的自适应杂交差分演化模拟退火算法

苏清华1,2,胡中波1,熊一能1   

  1. 1.孝感学院 数学系,湖北 孝感 432000
    2.华中科技大学 数学系,武汉 430074
  • 收稿日期:2009-02-24 修回日期:2009-04-07 出版日期:2010-08-11 发布日期:2010-08-11
  • 通讯作者: 苏清华

Cluster analysis based on self-adaptive hybrid differential evolution with simulated annealing algorithm

SU Qing-hua1,2,HU Zhong-bo1,XIONG Yi-neng1   

  1. 1.Department of Mathematics,Xiaogan University,Xiaogan,Hubei 432000,China
    2.Department of Mathematics,Huazhong University of Science and Technology,Wuhan 430074,China
  • Received:2009-02-24 Revised:2009-04-07 Online:2010-08-11 Published:2010-08-11
  • Contact: SU Qing-hua

摘要: 针对K-均值聚类算法对初始值敏感和易陷入局部最优的缺点,提出了一个基于自适应杂交差分演化模拟退火的K-均值聚类算法。该算法以差分演化算法为基础,通过模拟退火算法的更新策略来增强全局搜索能力,并运用自适应技术来选择学习策略、确定算法的关键参数。实验结果表明,该算法能较好地克服传统K-均值聚类算法的缺点,具有较好的全局收敛能力,且算法稳定性强、收敛速度快,将新算法与传统的K-均值聚类算法以及最近提出的几个同类聚类算法进行了比较。

关键词: 聚类分析, 差分演化算法, 模拟退火算法, 自适应技术, K-均值聚类算法

Abstract: The classical k-means clustering runs the risk of being trapped by local optima and its initial classical centers are difficulty in being set.In this paper,a novel k-means cluster analysis algorithm based on self-adaptive hybrid differential evolution with simulated annealing algorithm is proposed to overcome the disadvantage of the classical k-means algorithm.In the proposed algorithm,the choice of learning strategy and several critical control parameters are not required to be pre-specified.During evolution,the suitable learning strategy and parameters setting are gradually self-adapted according to the learning experience.With the aid of simulated annealing strategy,the proposed algorithm is able to improve the global search ability of conventional differential evolution algorithm.Numerical experiment results show that the new algorithms could overcome the faults of the classical k-means algorithm,and converge quickly.Comparative study exposes the two proposed algorithms as competitive algorithms for clustering.

Key words: cluster analysis, differential evolution algorithm, simulated annealing algorithm, self-adaptation, k-means cluster algorithm

中图分类号: