首页 | 本学科首页   官方微博 | 高级检索  
     

基于聚类和密度的KNN分类器训练样本约减方法
引用本文:艾英山,张德贤.基于聚类和密度的KNN分类器训练样本约减方法[J].计算机与数字工程,2009,37(5):10-12.
作者姓名:艾英山  张德贤
作者单位:河南工业大学信息科学与工程学院,郑州,450001
摘    要:提出了一种基于聚类和密度的KNN分类器训练样本约减方法。使用KNN分类器进行文本分类的时侯,由于训练样本在类别内分布的不均匀,会造成分类准确性的下降,而且相似度计算量非常大。新方法根据训练样本的密度采用聚类的方法,约减了一定数量的“噪声”样本。实验表明,使用该方法能同时提高KNN分类器的准确率和效率。

关 键 词:K近邻法  样本聚类  样本密度

A Method for Reducing the Amount of Training Samples in KNN Text Classification Based on Clustering and Density
Ai Yingshan,Zhang Dexian.A Method for Reducing the Amount of Training Samples in KNN Text Classification Based on Clustering and Density[J].Computer and Digital Engineering,2009,37(5):10-12.
Authors:Ai Yingshan  Zhang Dexian
Affiliation:College of Information Science and Engineering;Henan University of Technology;Zhengzhou 450001
Abstract:This paper forward a method for reducing the amount of training samples in KNN text classification based on clustering and density.The KNN classifier may decrease the precision of classification because of the uneven density of training samples.It has large computational demands when similarity is computed.In this paper,the new method reduces an amount of noisy samples based on clustering according to density of training samples.The test shows that this method can enhance precision and efficiency in KNN tex...
Keywords:KNN  samples clustering  density of samples  
本文献已被 CNKI 维普 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号