基于改进免疫算法的数据聚类策略研究  

Data clustering algorithm based on improved immune algorithm

作　　者：胡伟[1,2] 徐福缘[1] 

机构地区：[1]上海理工大学管理学院,上海200093 [2]绵阳师范学院数学与计算机科学学院,四川绵阳621000

出　　处：《计算机工程与设计》2012年第10期3940-3944,共5页Computer Engineering and Design

基　　金：国家自然科学基金项目(71171135);上海市(第三期)重点学科基金项目(S30504)

摘　　要：针对传统聚类算法聚类质量不够理想、自适应性不强和易陷入局部极小值等缺陷,提出一种基于改进免疫算法的数据聚类算法,该算法通过引入生物免疫系统中的精英保留策略和期望繁殖率,使适应度高的个体得到保留,浓度高的个体得到抑制,提高了算法的自适应性和精度,在后期通过利用混沌优化方法,使算法的局部搜索能力得到增强。实验结果表明,该算法比传统的聚类算法具有更好的性能。Aiming at the defect of traditional clustering algorithms for the poor quality of the clustering, week adaptivity and eas- ily trapping into local minima, an immune clustering algorithm based on the elitist strategy and chaos optimization is proposed. By introducing elitist strategy and expected rate of reproduction in the biological immune system, individual of high fitness is re tained and individual of high concentrations is inhibited. At the same time, the algorithm＇s self-adaptability and accuracy are im- proved. By taking advantage of the chaos optimization method in the late stage, the local search ability of the algorithm is strengthened. Experimental results show that the algorithm has better performance than the traditional clustering algorithm.

关 键 词：聚类 免疫算法 数据分析 混沌优化 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]