摘要
本文在克隆选择免疫算法和层次聚类的基础上,提出一种动态聚类算法。该算法无需先验知识,首先初始化与抗原相同规模的抗体,然后根据亲和力进行抗原识别、抗体抑制和合并,完成一轮聚类;再利用aiNET免疫网络模型动态确定聚类后的抗体的变异方向,实施强目的性变异,变异率反比例于进化代数动态调节,使变异后相似的抗体进一步合并,如此反复直到满足终止条件。仿真的实验结果表明,该算法比传统的聚类方法具有更好的聚类结果和更高的性能。
According to the basis of clonal selection immune algorithm and hierarchical clustering, an improved dynamic clustering algorithm is presented, in which no pre-knowledge is needed. Firstly the same size of antibodies as the antigens is initialized; Secondly antigen recognizing, antibody restraining and merging are performed based on antibody affinity to complete a round of clustering;Thirdly, in order to do some motivated mutating, the mutating location of antibodies is determined using aiNET immune network model and the mutating rate is dynamically adjusted inversely proportional to the generation count of immune evolution. After dynamic mutating, the similar antibodies are merged again. Then it repeats the above processes until meets the ending condition. Experimental result shows that it has better clustering results and performance than traditional ones.
出处
《微计算机信息》
北大核心
2007年第27期255-257,共3页
Control & Automation
关键词
聚类
克隆选择
免疫算法
变异
clustering, clonal selection, immune algorithm, mutating
