CAJ | 학술논문

针对传统K-均值聚类算法对初始中心选择敏感和全局搜索能力不足的缺点,提出一种结合灰狼优化和K-均值的混合聚类算法 (GWO-KM ). 将灰狼优化智能算法首次应用到聚类分析领域,利用灰狼算法良好的勘探能力去寻找使聚类结果最佳的一组聚类中心,克服了原始聚类算法对初始中心点的过度依赖. 通过在UCI标准数据集上的仿真实验表明, 该混合聚类算法相对于传统的K-均值聚类算法和其他改进算法,在收敛速度、聚类质量和稳定性上都表现更佳.
침대전통K-균치취류산법대초시중심선택민감화전국수색능력불족적결점,제출일충결합회랑우화화K-균치적혼합취류산법 (GWO-KM ). 장회랑우화지능산법수차응용도취류분석영역,이용회랑산법량호적감탐능력거심조사취류결과최가적일조취류중심,극복료원시취류산법대초시중심점적과도의뢰. 통과재UCI표준수거집상적방진실험표명, 해혼합취류산법상대우전통적K-균치취류산법화기타개진산법,재수렴속도、취류질량화은정성상도표현경가.
Considering the disadvantages of K-means Cluster Algorithm, such as inadequacy for global search and being sensitive to initial cluster centers, this paper proposes a hybrid clustering algorithm based on Grey Wolf Optimizer and K-means (GWO-KM). Grey Wolf Optimizer is applied in the field of cluster analysis for the first time. With its good ability of exploration, the new intelligent algorithm helps K-means cluster find a set of cluster centers which can get the best clustering result, thus avoiding the original algorithm's over-dependence on initial centers. The experiments based on UCI show that, the hybrid algorithm can result in faster convergence speed, higher accuracy, and greater stability, compared with traditional k-means algorithm and other improved algorithms.