摘要
针对固定网格划分技术存在的维度可扩展性差,而自适应网格划分技术未充分考虑数据集分布特征等问题,提出了一种基于概率统计理论的自适应网格聚类算法.采用概率统计和图覆盖技术,且能识别任意形状和大小的聚类,时间复杂度是数据集大小和数据维度的线性函数.实验结果表明该聚类是有效的.
In order to solve the problems that low expansibility of dimension existented in fixed grids partitioning technique while distributing feature of data .set disconsidered in auto-adapted grid clustering algorithm sufficiently, an auto-adapted grid clustering algorithm based on probability Star. is propesed （auto-adapted grid clustering algorithm based on probability Star. ,AGAR）. The text adopt probability Star. Technique and graph-based overlay technique, which can discover arbitrary shapes and sizes of clusters, and the time complexity is linear to the size of the input data set or data dimensions. The experimental results show that the algorithm is effective.
出处
《微电子学与计算机》
CSCD
北大核心
2008年第5期173-175,178,共4页
Microelectronics & Computer
基金
国家科技支持计划课题(2006BAF01A18)
关键词
聚类
自适应网格
可扩展性
clustering
auto-adapted grid
expansibility
