Logo image
A novel hybrid clustering approach based on K-harmonic means using robust design
期刊文章   同儕審查

A novel hybrid clustering approach based on K-harmonic means using robust design

Wei-Chang Yeh, Chyh-Ming LaiKuei-Hu Chang
Neurocomputing, 卷.173, 頁碼.1720-1732
01/2016

摘要

Data clustering K-harmonic means Particle swarm optimization Simplified swarm optimization Taguchi method Computer Science Applications Cognitive Neuroscience Artificial Intelligence
The K-harmonic means (KHM) algorithm has been proposed for solving the initialization problem of K-means recently. However, the KHM still suffers from being trapped into local optima. In order to solve this problem, this paper presents a novel hybrid algorithm named RMSSOKHM based on KHM and a modified simplified swarm optimization. The proposed RMSSOKHM adopts the rapid centralized strategy (RCS) to increase the convergence speed and the minimum movement strategy (MMS) to effectively and efficiently search better solutions without trapping in local optima. In addition, the parameter settings of the proposed approach were optimized by using the Taguchi method. The performance of the proposed RMSSOKHM was examined and compared with the existing-known methods using eight benchmark datasets. The experimental results indicated that the proposed RMSSOKHM is superior to its competitors in terms of the quality of solutions and the efficiency of performance.

相關連結

指標

1 檢視次數

詳細資料

Logo image