| 一种基于最短距离聚类的K最近邻分类算法 |
| |
| 引用本文： | 陈江丽,张嵘.一种基于最短距离聚类的K最近邻分类算法[J].新乡学院学报(自然科学版),2014(12):29-33. |
| |
| 作者姓名： | 陈江丽 张嵘 |
| |
| 作者单位： | 临沧师范高等专科学校信息科学与技术系 |
| |
| 基金项目： | 云南省教育厅科学研究基金项目(2013C037);临沧师范高等专科学校科学研究基金项目(LCSZL2013009) |
| |
| 摘 要： | 针对传统K最近邻（KNN）分类法执行效率低的问题,提出一种改进的K最近邻分类法。先采用最短距离聚类法分别对训练样本和测试样本进行聚类,生成一些小簇和孤立点,再对小簇或孤立点使用改进的K最近邻方法进行分类。改进后的方法能极大地缩小分类样本的规模,降低计算成本,提高分类效率。
|
| 关 键 词： | K最近邻分类 训练样本 测试样本 聚类 最短距离 |
A K Nearest-neighbor Classification Algorithm Based on Shortest Distance Clustering |
| |
| Authors: | CHEN Jiangli;ZHANG Rong |
| |
| Institution: | CHEN Jiangli;ZHANG Rong;Department of Information Science and Technology,Lincang Teachers’ College; |
| |
| Abstract: | Aiming at low efficiency of traditional K nearest-neighbor（KNN）classification method,this paper proposed an improved K nearest-neighbor classification algorithm.At first training samples and test samples were clustered using the shortest distance clustering method,and some clusters and isolation were generated.Then clusters or isolation were classified by the improved K nearest-neighbor classification method.The improved method can reduce the size of sample greatly,reduce the calculation expense,and improve the efficiency of classification. |
| |
| Keywords: | K nearest-neighbor classification training sample test samples cluster the shortest distance |
| 本文献已被 CNKI 维普 等数据库收录！ |
