一种新的基于进化计算的聚类算法  

New clustering algorithm based on evolutionary computation

作　　者：张俊溪[1] 吴晓军[2] 

机构地区：[1]西安航空技术高等专科学校生物医电工程学院,西安710077 [2]西北工业大学自动化学院,西安710072

出　　处：《计算机工程与应用》2011年第24期111-114,共4页Computer Engineering and Applications

基　　金：国家自然科学基金(No.10702065)~~

摘　　要：聚类是数据挖掘领域的重要研究内容之一。针对遗传聚类算法较好的稳定性与粒子群优化算法较强的局部搜索能力,在交叉、变异算子后叠加粒子群优化算子的方法实现了二者的结合,提出了GAPSO聚类算法,既保持了遗传算法的稳定性与泛化性的优势,又发挥了PSO算法收敛效率高的特点。通过对10组二维空间上的聚类样本进行实验研究显示,GAPSO聚类算法在收敛效率上显著优于GA聚类算法,在稳定性上优于PSO聚类算法。Cluster analysis which plays an important role in data mining,is widely used.It has important value both in theo-ry and application.Considering the stability of the genetic algorithm and the local searching capability of particle swarm opti-mization in clustering,the two algorithms are combined.Particle swarm optimization operators are implemented after the cross-over and mutation operators,and GAPSO clustering algorithm is put forwarded.Simulation results are given to illustrate the stability and convergence of the proposed method.GAPSO is proved to be easier to carry out,faster to converge and more stable than other methods.

关 键 词：数据挖掘 聚类 遗传算法 粒子群优化算法 遗传粒子群优化算法(GAPSO) 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]