A novel multiobjective K-harmonic means clustering algorithm using Levy Flight Cuckoo Search

作者：Song Anping^*; Bai Xuebin; Ding Xuehai; Zhang Wu

来源：Journal of Computational Information Systems, 2013, 9(24): 9953-9964.

DOI：10.12733/jcis8427

• Cluster validity indices
• Cuckoo searches
• K-harmonic Means
• K-harmonic means clustering
• Levy flights
• Meta heuristic algorithm
• Multi-objective clustering
• Objective functions

K-harmonic means (KHM) is a centroid based clustering algorithm, which will easily converge to local optimal and optimize single objective function. Recent researches have shown that there is no single cluster validity index works equally well for Diffierent kinds of datasets. Multiobjective clustering is used to solve this problem, which optimizes multiple validity measures simultaneously. Besides Levy ight Cuckoo Search is a recent developed nature inspired meta-heuristic algorithm that works efficiently for clustering. These facts motivate us to propose a novel multiobjective KHM clustering algorithm using Levy Flight Cuckoo Search (MOKHMCS) which optimizes the KHM objective function and Xie-Beni index simultaneously. Some modifications have been made in Cuckoo Search to apply to MOKHMCS. It will produce a near-Pareto optimal non-dominated set of solutions in the final generation and PBMF index is used to select a final solution in the set. The experimental results on two artificial and UCI datasets and three bioinformatics datasets indicate the superiority of MOKHMCS compared to KHM, Cuckoo Search based KHM and Particle Swarm Optimization based multiobjective KHM. The results also show the effectiveness of multiobjective clustering and the efficiency of Levy ight Cuckoo Search helps KHM escape from local optimal and speed up convergence.

• 出版日期2013
• 单位上海大学