• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

计算机工程与科学

• 人工智能与数据挖掘 • 上一篇    下一篇

基于二次网格优化的粒子群模糊聚类算法

汪赫瑜,唐敏影,任建华   

  1. (辽宁工程技术大学电子与信息工程学院,辽宁 葫芦岛 125105)
  • 收稿日期:2017-11-27 修回日期:2018-01-11 出版日期:2019-02-25 发布日期:2019-02-25

A particle swarm fuzzy clustering algorithm
 based on quadratic grid optimization

WANG Heyu,TANG Minying,REN Jianhua   

  1. (School of Electronics and Information Engineering,Liaoning Technical University,Huludao 125105,China)
  • Received:2017-11-27 Revised:2018-01-11 Online:2019-02-25 Published:2019-02-25

摘要:

针对模糊C均值聚类算法易受初始聚类中心影响且收敛速度慢的缺陷,提出一种基于二次网格优化的粒子群模糊聚类算法GridPFcm。该算法首先将数据空间网格化,依据深度优先遍历规则,连通相邻密集网格单元,计算连通网格的相对密度,选取相对密度值最大的连通网格,计算中心位置,初始化聚类中心。然后,按照基于网格空间的单维向量变化原理,实现最佳粒子全局寻优,进一步优化初始聚类中心,以降低初始聚类中心选取对聚类效果的影响度。最后,通过实验表明,该算法能够加快寻优收敛速度,提高聚类效率和效果。
 

关键词: 模糊C均值聚类, 连通网格, 相对密度, 中心位置, 粒子

Abstract:

The fuzzy C-means clustering algorithm is susceptible to the initial clustering center and the convergence rate is slow. We present a particle swarm fuzzy clustering algorithm based on quadratic grid optimization. The algorithm firstly grids the data space, and connects adjacent dense grid cells according to the depth-first traversal rule, calculates the relative density, selects the connected grid with highest relative density value, calculates its central location, and initializes the clustering center. Then, according to the principle of single-dimensional vector variation based on grid space, it achieves the optimal particle global optimization and further optimizes the initial clustering center to reduce the effect on clustering. Experimental results show that the algorithm can accelerate convergence speed and improve the clustering efficiency and effect.
 

Key words: fuzzy C-means clustering, connected grid, relative density, central location, particle