计算机工程与应用 ›› 2012, Vol. 48 ›› Issue (36): 186-189.

• 图形、图像、模式识别 • 上一篇    下一篇

SA-BFSN:一种自适应基于密度聚类的算法

陈  昊,侯慧群,杨承志,邱  磊   

  1. 空军航空大学,长春 130022
  • 出版日期:2012-12-21 发布日期:2012-12-21

SA-BFSN:adaptive algorithm based on density clustering

CHEN Hao, HOU Huiqun, YANG Chengzhi, QIU Lei   

  1. Aviation University of Air Force, Changchun 130022, China
  • Online:2012-12-21 Published:2012-12-21

摘要: 针对BFSN算法需要人工输入参数r和[λ]的缺陷,提出了一种自适应确定r和[λ]的SA-BFSN聚类方法。该方法通过Inverse Gaussian拟合判断r参数,通过分析噪声点数量的分布特征选择合适的[λ]值。算法测试表明,使用SA-BFSN无需人工输入参数,能够实现聚类过程的全自动化,能够有效处理任意形状、大小和密度的簇。

关键词: 数据挖掘, 密度聚类, 基于广度优先搜索邻居的聚类算法(BFSN), 自适应基于广度优先搜索邻居的聚类算法(SA-BFSN)

Abstract: Algorithm for BFSN defects that require manual input parameters r and [λ], an adaptive SA-BFSN clustering method that can automatically determine r and [λ] is proposed. The method determines r by Inverse Gaussian fitting parameters, and by analyzing the distribution of the number of noise points to select the appropriate value of [λ]. Algorithm tests show that use of SA-BFSN doesn’t need human input parameters, to achieve full automation of the clustering process, to deal effectively with any shape, size and density of the cluster.

Key words: data mining, density clustering, Broad First Search Neighbors(BFSN), Self-Adaptive Broad First Search Neighbors(SA-BFSN)