基于距离浓度的K-均值聚类算法

提出的基于距离浓度的K-均值聚类算法把聚类的数据对象视为抗原,聚类中心看作是免疫系统中的抗体,聚类过程表示为免疫系统不断产生抗体,识别抗原,最后产生出可以捕获抗原的最佳抗体过程.定义了抗体浓度和亲和度,使得抗体之间的距离越大,其距离浓度越小,反之则浓度越大,从而提高了算法的搜索效率.设计了抗体的期望繁殖率计算方法和克隆变异方法.仿真结果表明:该算法不仅克服了传统的K-均值聚类算法易陷入局部极小值的缺点,而且避免了对初始化选值敏感性的问题,同时也有较快的收敛速度.

K-means clustering algorithm based distance concentration

Using the immune recognizing principle,the data object to cluster was denoted as the antigens set,and the clustering center was the antibodies set.The clustering was the process to obtain the best antibodies to catch the antigens by producing the antibodies and recognizing the antigens unceasingly.The distance concentration and the affinity,between antibody and antigen,and between antibody and antibody,were defined about the K-means clustering.It makes that the longer the distance between antibodies is,the lower the distance concentration is,per contra.This design improves the clustering search efficiency.The antibody reproduction function was also proposed,and the antibody cloning algorithm was presented simultaneously.The experimental results show that the algorithm not only avoids the local optima and is robust to initialization,but also increases the convergence speed.