Hybrid Iterative K-Means Clustering with Improved Moth-Flame Optimization

被引：0
|
作者：
Huang H. [1 ,2 ]
Li X. [1 ]
Wu K. [1 ]
Guo L. [3 ]
Wang H. [1 ]
Ru F. [1 ]
机构：
[1] School of Electronic Control, Chang'an University, Xi'an
[2] Shaanxi Road Traffic Intelligent Detection and Equipment Engineering Technology Research Center, Xi'an
[3] UAV National Engineering Research Center, Northwestern Polytechnical University, Xi'an
来源：
Hsi-An Chiao Tung Ta Hsueh/Journal of Xi'an Jiaotong University | 2020年 / 54卷 / 09期
关键词：
Average distance category; Cluster center; K-means clustering; Maximum and minimum distance product; Moth-flame optimization;
D O I：
10.7652/xjtuxb202009003
中图分类号：
学科分类号：
摘要：
A new K-means clustering algorithm with improved moth-flame optimization is proposed to solve the problem that the current K-means clustering (KMC) algorithm has great randomness in selecting the initial clustering center, poor global search ability and low clustering accuracy. The maximum and minimum distance function is adopted to initialize the clustering center to avoid the problem that KMC algorithm is sensitive to the random initial clustering center. Then the moth-flame optimization is improved by spline interpolation to heighten the convergence rate and optimization accuracy. The average distance category is taken as the fitness function to guide interpolation moth-flame optimization to optimize the clustering center in the process of KMC iteration so as to improve the clustering accuracy. Compared with KMC algorithm, K-means++ algorithm and fuzzy c-means clustering algorithm on the international standard data sets Iris, Wine and Seeds, the experimental results show that the IMFO-KMC algorithm achieves the most significant improvement on the Iris data set. Compared with the other algorithms, the accuracy of IMFO-KMC algorithm is improved by 0.67%-4.18%, and the normalized mutual information is improved by 1.5%-4.01%. © 2020, Editorial Office of Journal of Xi'an Jiaotong University. All right reserved.
引用
收藏
页码：32 / 39
页数：7
相关论文
共 21 条
  • [1] YU Q Y, LUO Y L, CHEN C M, Et al., Outlier-eliminated k-means clustering algorithm based on differential privacy preservation, Applied Intelligence, 45, 4, pp. 1179-1191, (2016)
  • [2] PATEL G K, DABHI V K, PRAJAPATI H B., Clustering using a combination of particle swarm optimization and K-means, Journal of Intelligent Systems, 26, 3, pp. 457-469, (2017)
  • [3] CHEN Xingshu, WU Xiaosong, WANG Wenxian, Et al., An improved initial cluster centers selection algorithm for K-means based on features correlative degree, Journal of Sichuan University (Engineering Science Edition), 47, 1, pp. 13-19, (2015)
  • [4] XIAO Yang, LI Ping, WANG Peng, Et al., An adaptive K-means initialization method based on minimum deviation, Journal of Changchun University of Science and Technology (Natural Science Edition), 38, 5, pp. 140-144, (2015)
  • [5] ZHENG Dan, WANG Qianping, Selection algorithm for K-means initial clustering center, Journal of Computer Applications, 32, 8, pp. 2186-2188, (2012)
  • [6] ZHANG Qiang, WANG Hongwei, CHEN You, Et al., Rough fuzzy c-means clustering algorithm based on self-adaptive weights, Microelectronics & Computer, 33, 12, pp. 80-84, (2016)
  • [7] YU Jinxia, CAI Zixing, DUAN Zhuohua, An improved genetic clustering algorithm for feature extraction of laser scanner, Pattern Recognition and Artificial Intelligence, 21, 4, pp. 535-540, (2008)
  • [8] YU Jinping, ZHENG Jie, MEI Hongbiao, K-means clustering algorithm based on improved artificial bee colony algorithm, Journal of Computer Applications, 34, 4, pp. 1065-1069, (2014)
  • [9] MIRJALILI S., Moth-flame optimization algorithm: a novel nature-inspired heuristic paradigm, Knowledge-Based Systems, 89, pp. 228-249, (2015)
  • [10] KHALILPOURAZARI S, PASANDIDEH S H R., Multi-item EOQ model with nonlinear unit holding cost and partial backordering: moth-flame optimization algorithm, Journal of Industrial and Production Engineering, 34, 1, pp. 42-51, (2017)
123