基于代理模型的分布式聚类算法  

RFID-based vehicle management system

作　　者：田野[1] 彭利标[1] 

机构地区：[1]天津理工大学中环信息学院,天津300380

出　　处：《电子设计工程》2013年第17期7-10,13,共5页Electronic Design Engineering

摘　　要：II DDM模型是现有的分布式聚类模型中性能较好的一种个体合作以及串行工作方式固有的不足,在该模型基础上引入分层的思想,提出了一种新的分布式聚类算法模型HII DDM,该模型将分布式数据挖掘系统分解成小的子系统,并通过代理层层向上聚类,与单纯使用II DDM模型的系统比较比较,有更好的伸缩性和执行效率,通信代价也有所降低,特别适合数据库和站点量大的数据聚类问题。With the introduction of level design it comes a hierarchical design model HIIDDM based on distributed data mining,using divide and conquer strategy to decompose a large system into several smaller subsystem,which levels the use of incremental mining model integration tasks.It proposes a new distributed clustering algorithm HIIDC according to hierarchical clustering algorithm.The experiments show that simply using the clustering algorithm II DDM model comparison,the algorithm has more good flexibility and efficiency in the implementation,communication.Costs are greatly reduced,especially for large-scale data clustering.

关 键 词：数据挖掘 分布式聚类 代理模型 层次增量 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]