基于云关联规则的蚁群聚类算法研究

摘要

Based on the excellent features of cloud model in the non-normative knowledge representation, the paper presents an algorithm which improves ant colony clustering on cloud-model association rules. By the dynamic and quick soft partition of concept based on cloud-model association rules in neighborhood, the greatest cohesive core is produced, then the calculating formula of accepted scores is redefined and normal membership clouds on the domain of membership are smoothly changed and the clustering-convergence process is accelerated. Experimental results show that the algorithm can form the clustering of high purity and improve the convergence speed of clustering process.%基于云模型在非规范知识表示中的优良特征,本文提出了一种基于云关联规则的改进蚁群聚类算法.通过在邻域内进行基于云模型关联规则的概念快速动态软划分来产生最大内聚核,重新定义接受分数模型,使属性论城上正态隶属云实现平滑变迁,加快了聚类收敛过程.实验结果表明,改进算法能产生高纯度的聚类簇,提高了聚类过程的收敛速度.