基于改进HHO与K-Medoids的混合聚类算法  

Hybrid clustering algorithm based on improved HHO and K-Medoids

作　　者：李姣 王秋萍[1] 戴芳[1] LI Jiao;WANG Qiuping;DAI Fang(Faculty of Sciences,Xi’an University of Technology,Xi’an 710054,China)

机构地区：[1]西安理工大学理学院,陕西西安710054

出　　处：《西安理工大学学报》2022年第3期410-420,共11页Journal of Xi'an University of Technology

基　　金：国家自然科学基金资助项目(61976176)。

摘　　要：针对K-Means在聚类过程中对离群点敏感以及容易陷入局部最优的不足,本文提出一种基于改进HHO(IHHO)与K-Medoids的混合聚类算法(IHHO-KMedoids)。在IHHO中,带有Logistic混沌扰动的控制参数策略更好地实现了探索与开发之间的平衡,集成变异策略提高了算法的全局搜索能力,翻筋斗觅食策略增强了种群多样性,避免算法陷入局部最优。将所提IHHO与5种其他群智能算法和4种改进的HHO算法在CEC 2014测试函数上进行对比,实验结果表明IHHO算法的优化效果较好,求解精度较高。K-Medoids与K-Means相比对噪声点和离群点更鲁棒。IHHO-KMedoids算法稳定性好,不易陷入局部最优。UCI数据集和文本数据集上的仿真结果表明IHHO-KMedoids算法效率高,聚类精度高。A hybrid clustering algorithm(IHHO-KMedoids) based on the improved HHO(IHHO) and K-Medoids is proposed in this paper for solving the issues of K-Means which is sensitive to outliers and easy to fall into local optimum. In IHHO, the control parameter strategy with Logistic chaotic disturbance better achieves the balance between exploration and exploitation, the ensemble mutation strategy improves the global search ability of the algorithm, and the somersault foraging strategy increases the diversity of the population and avoids the algorithm falling into local optimum. The proposed IHHO is compared with five other swarm intelligence algorithms and four improved HHO algorithms on the CEC 2014 benchmark functions with the experimental results showing that IHHO has better optimization ability and higher accuracy. K-Medoids is better robust against noise and outliers compared with K-Means. IHHO-KMedoids algorithm has high stability and is not easy to fall into local optimum. The simulation results on UCI datasets and a text dataset show that the IHHO-KMedoids algorithm has higher efficiency and clustering accuracy compared with contrastive algorithms.

关 键 词：Harris鹰优化算法 LOGISTIC映射 集成变异策略 翻筋斗觅食策略 K-Medoids算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]