基于多染色体演化的自适应类别数聚类方法  

Count adaptive clustering algorithm based on multiple-chromosome evolution

作　　者：倪广翼[1] 章孝灿[1] 苏程[1] 俞伟斌[1] 

机构地区：[1]浙江大学地球科学系,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2014年第6期980-986,1023,共8页Journal of Zhejiang University：Engineering Science

摘　　要：为了解决在遗传算法聚类分析中影响算法效率的互相关性问题以及在没有先验知识的情况下确定类别数问题,在充分分析基因的互相关性对算法效率和收敛性影响的基础上,借鉴多染色体生物的进化特性,提出多染色体取代传统单染色体的遗传算法.算法在进化过程中充分利用类簇之间的相互关系,提高了遗传算法的效率和收敛能力,并且在遗传过程中类别数量可变;为了明确地控制类别数,采用基于分布拟合的适应度函数,为在没有先验知识的情况下确定类别数提供了分析依据.通过与K均值的遗传算法(KGA)、最大期望算法(EM算法)的对比分析以及针对遥感影像的实验表明,该遗传算法在对类别数能进行自适应控制的基础上,在效率和收敛性上也都能取得较好的效果.Two well--known problems exist when genetic algorithms （GAs） applied to data clustering： the ＂content-sensitivity＂ problem which restrains efficiency, and how to determine the count of clusters without prior knowledge. In order to resolve these problems, by fully researching on the nature of how ＂context-sensitivity＂ affects efficiency and convergency, a multiple chromosome genetic clustering algorithm was proposed, which takes the advantage of the evolvement of multiple chromosome creatures. The algorithm improved the efficiency and convergency, by making use of the interrelationship among all clusters while evolving, moreover, by this way the count of cluster is variable in each individual. To control the count of clusters, fitness functions which based on distribution fitting was defined, which can provide decision information without prior knowledge. The comparison of the algorithm with the K genetic algorithm （KGA） and the expectation-maximization （EM） algorithm and the results of experiments on remote sensing image show promising results on efficiency and convergency with cluster count automatically adjusted.

关 键 词：聚类分析 遗传算法 多染色体 适应度 类别数 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]