基于人工免疫网络的动态聚类算法.PDF

第8期 电 子 学 报 Vol.32 No. 8 2004 年8 月 ACTA ELECTRONICA SINICA Au . 2004 基于人工免疫网络的动态聚类算法 钟 将, 吴中福, 吴开贵, 欧 灵 ( 重庆大学计算机学院, 重庆400044) : 聚类分析的两个基本任务是分析数 集中簇的数量以及这些簇的位置. 大多数的聚类方法通常只关注 后一个问题. 为了在聚类数不确定的情况下实现聚类分析, 本文提出了一种新的结合人工免疫网络和遗传算法的动态 聚类算法) DCBIG. 新算法主要包含两个阶段: 先使用人工免疫网络算法获得聚类可行解, 然后使用遗传算法依 聚 类可行解实现动态聚类. 本文对获得聚类可行解的条件和概率进行了分析. 仿真实验结果表明与现有方法相比, 新方 法具有更高的收敛概率和收敛速度. : 动态聚类; 免疫网络; 聚类可行解; 收敛 : TP183 : A : 037222112 (2004) 082 12682 05 A Novel Dynamic Clustering Algorithm Base d on Artificial Immune Network ZHONG Jian , WU Zhon 2fu, WU Kai2 ui, OU Lin ( Collegeof Computer Science and Engineering, Chongqing University, Chongqing 400044, China) Ab stract: Cluster analysis aims at answerin two main questions:how many clusters there are in the data set and where they are located.Usually,the traditional clusterin al orithms only focus on the last problem. In order to solve the two problems at the same time,this paper proposes a novel dynamic clusterin al orithm called DCBIG,which is based on the immune network and enetic al o2 rithm. The al orithm includes twophases,be ins by runnin immune network al orithm to find a feasible solution, and then employs e2 netic al orithm to search the optimum number of clusters and the location of each cluster accordin to the feasible solution.Also,the probabilities and the conditions to acquire a feasible solution throu h immune network al orithm are discussed in this paper.