基于粒度原理的蚁群聚类算法

Ant Colony Clustering Algorithm Based on Principle of Granularity

在过去10多年中,蚁群算法(AC)的研究和应用取得了很大的进展,大量结果证明了算法的有效性和在某些领域的优势。文章从信息粒度的角度出发,解决了传统聚类算法中对样本“抱团”性质的客观描述和分类算法中分类专家主观先验知识之间的不协调性。并将蚁群系统模型引入聚类模型中,提出了一种基于粒度原理的蚁群聚类新方法。仿真结果表明上述方法是可行和有效的。

The researches and applications on ant colony algorithm have made great progresses in the past ten years or more. A number of results prove the validity of the algorithm and its advantages in some fields. From the view of information granularity, this article solves the problem that there is no correspondence between the objective description of the congregation quality of the examples in traditional clustering algorithm and the subjective knowledge of the classification experts in the classification algorithm. And by introducing the ant colony system into the clustering model, it presents a new kind of ant colony clustering algorithm. The simulation results demonstrate that the above approach is reasonable and efficient.