一种改进的蚁群聚类分析算法  

An Improved Ant Clustering Analysis Algorithm

作　　者：甘泉[1] 王慧[2] 

机构地区：[1]平顶山学院计算机与科学技术学院,河南平顶山467000 [2]平顶山学院招生就业处,河南平顶山467000

出　　处：《系统仿真技术》2015年第3期219-223,206,共6页System Simulation Technology

基　　金：河南省科技厅攻关项目(KJT142102210226)

摘　　要：提出了一种改进的蚁群聚类分析算法。该算法改进了经典的LF算法。利用短期记忆和网格信息素的局部分布控制蚂蚁的随机移动,利用信息熵作为蚂蚁运动状态转换规则,通过对信息熵的计算与比较,更改了数据对象拾起和放下的判断规则。设置的参数减少了,不仅每次放下对象时能够减少小块区域的信息熵,拾起时能够增加小块区域的信息熵,还能加快聚类过程,达到好的聚类结果。结果表明,该算法显示出了较高的稳定性和准确率。An improved ant colony clustering algorithm is developed. This algorithm improved classic LF algorithm by employing short-term memory and local distribution grid to control ant pheromone random movement, information entropy state of motion to act as ants＇ conversion rules. The judgment rule of picking up and dropping data is changed based on calculating and comparing the information entropy. This reduces the set parameters. Not only the entropy of small area which putting down the object reduces, but also the entropy of small area which picking up the object increases. The method speeds up of process of clustering and achieves good clustering results. These results prove that the algorithm has high stability and accuracy.

关 键 词：聚类 蚁群聚类 信息素 信息熵 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]