摘要
针对SCMD算法存在的两大不足提出了改进,改进的半监督聚类算法在原算法的基础上添加对两种问题的处理,问题一的解决方法是查找可能会丢失的簇,添加Eps,以解决先验约束不充分时不能检测到所有的簇;问题二的解决方法是分配边界簇,以解决簇内多密度问题。实验证明SCMDFC算法在处理多密度数据集时具有良好的聚类质量。
At proposing improvements of the two existing deficiencies of SCMD algorithm.Improved semi-supervised clustering algorithm adds processing of two kinds problem based on the original algorithm.Solution of the first problem is in search of the cluster that can be lost and adding the Eps in order to solve the problem of hardly fully detecting all the cluster in the condition of insufficiency of a priori constraint.Solution of the second problem is to allocate boundaries to solve the problem of cluster-heads multidimensional.Experiment proofs that SCMDFC algorithm has better clustering quality in dealing with a multidimensional data set.
出处
《网络安全技术与应用》
2014年第4期85-86,共2页
Network Security Technology & Application
