计算机工程与应用 ›› 2023, Vol. 59 ›› Issue (17): 91-98.DOI: 10.3778/j.issn.1002-8331.2205-0436

• 理论与研发 • 上一篇    下一篇

结合柯西分布和蚁狮算法改进的模糊聚类算法

吴辰文,王莎莎,曹雪同   

  1. 兰州交通大学 电子与信息工程学院,兰州 730070
  • 出版日期:2023-09-01 发布日期:2023-09-01

Fuzzy Clustering Algorithm Combined with Cauchy Distribution and Ant Lion Algorithm

WU Chenwen, WANG Shasha, CAO Xuetong   

  1. School of Electronics and Information Engineering, Lanzhou Jiaotong University, Lanzhou 730070, China
  • Online:2023-09-01 Published:2023-09-01

摘要: 针对模糊聚类对初始聚类中心依赖性较强且易陷入局部最优解的问题,提出了一种结合柯西分布和蚁狮算法改进的模糊聚类算法(CALOFCM)。引入柯西分布函数变异蚁狮算法,使得个体受局部极值点的约束力下降,从而增加跳出局部最优解的概率。使用优化后的蚁狮算法生成的精英蚁狮作为模糊[C]均值(fuzzy C-means,FCM)算法的初始聚类中心。分别在人工数据集和UCI数据集上进行了实验验证,并与K-means、DBSCAN、FCM、ALOFCM算法以及提出的算法进行实验对比。结果表明改进的算法获得了较好的聚类结果且在准确率、调整兰德系数和标准化互信息等评价指标上具有良好的聚类性能。

关键词: 数据挖掘, 模糊[C]均值算法, 蚁狮算法, 柯西分布

Abstract: Aiming at the problem that fuzzy clustering depends strongly on the initial clustering centers and easy to fall into local optimal solutions, a fuzzy clustering algorithm combined with Cauchy distribution and ant lion algorithm (CALOFCM) is proposed. The Cauchy distribution function variant ant lion optimization algorithm is introduced, which reduces the binding force of individuals by local extreme points, thus increasing the probability of escaping from the local optimum. The elite ant lions generated by the optimized ant lion algorithm are used as the initial clustering centers of the fuzzy C-means (FCM) algorithm. The comparison experiments of artificial data sets and UCI data sets show that compared with K-means, DBSCAN, FCM, ALOFCM algorithm, the proposed algorithm obtains better clustering effect and has good clustering performance in accuracy, adjusted rand index and normalized mutual information.

Key words: data mining, fuzzy C-means(FCM), ant lion optimization algorithm(ALO), Cauchy distribution