摘要
基于构造型神经网络引入一种新的数据流聚类相似性函数,并根据滑动窗口模型数据流聚类的特点,定义了平均覆盖和重叠覆盖等概念,进而提出基于构造型神经网络的滑动窗口模型数据流聚类算法.该算法可以降低计算量,提高聚类速度.大规模无线电监测数据聚类实验验证了该算法的有效性.
The article introduces a new clustering similarity based on the constructive neural networks.The conceptions of average cover and over cover are defined to meet the need of sliding windows model data stream clustering,and then a clustering algorithm for sliding windows model data stream is present based on the constructive neural networks.It can be found that the calculation can be lessened and the speed of clustering can be faster with the use of this algorithm.The experimental results of large-scale communication datasets demonstrate its efficiency.
出处
《小型微型计算机系统》
CSCD
北大核心
2010年第12期2355-2358,共4页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(60675031)资助
关键词
数据流
聚类
构造型神经网络
滑动窗口
data stream
clustering
constructive neural network
sliding window
