| 一种有效的密度有偏采样算法 |
| |
| 引用本文： | 于华. 一种有效的密度有偏采样算法[J]. 计算机应用与软件, 2008, 25(6): 247-249 |
| |
| 作者姓名： | 于华 |
| |
| 作者单位： | 宁波工程学院电子与信息工程学院,浙江,宁波,315016 |
| |
| 摘 要： | 密度有偏采样技术是针对不一致数据集提出的一种采样技术,现有密度有偏采样方法在寻找聚类时容易受异常点的影响而产生错误的聚类.基于这一缺陷,提出了一种新的密度有偏采样--DOSP,在寻找聚类时,该算法可以利用密度有偏采样存在的空间划分,在计算子空间密度的同时搜寻异常子空间,并屏蔽对异常子空间的采样,从而有效地消除异常点的影响.实验证明该算法不仅有效地消除了异常点的干扰,而且在高噪声情况下,算法的计算代价明显降低.
|
| 关 键 词： | 数据挖掘 聚类 密度有偏采样 异常点 |
| 修稿时间： | 2006-10-30 |
AN EFFICIENT DENSITY-BIASED SAMPLING ALGORITHM |
| |
| Yu Hua. AN EFFICIENT DENSITY-BIASED SAMPLING ALGORITHM[J]. Computer Applications and Software, 2008, 25(6): 247-249 |
| |
| Authors: | Yu Hua |
| |
| Affiliation: | Yu Hua(Electronic , Information College,Ningbo University of Technology,Ningbo 315016,Zhejiang,China) |
| |
| Abstract: | Density biased sampling(DBS)is a sampling technique specially presented for inconsistency dataset,but existing DBS methods are likely to have error clusters occur due to the impacts from outliers when searching clusters.In this paper,a new density biased sampling algorithm called DOSP(Deleting Outlier Selective Pass)is presented against this drawback.This algorithm,when searching clusters,resorting to existed partition of sections on density biased sampling,searches outlier sections during computing the den... |
| |
| Keywords: | Data mining Clustering Density biased sampling Outlier |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
