摘要
本文介绍了分类中常用的算法之一K-近邻法,分析了此分类算法相对于其他分类算法的优势及不足,并针对此方法在各类训练样本分布不等时造成的分类效率下降的问题作了相应的改进,提出了一种在相似度计算时能体现各类代表度的K-近邻法分类方法,减小了误判率.实验证明了所做的改进是有效的.
This paper introduces the K - nearest neighbor method, which is one of the commonly used algo- rithms in classification. The advantages and disadvantages of this classification algorithm compared with other clas- sification algorithms are analyzed. This method used for the problem of falling classification efficiency is made cor- responding improvementwhen the distribution of all kinds of training samples is unequal. And a K -nearest neigh- bor classification method is proposed, with which can reflect various kinds of representative degree when we com- pute similarity.
出处
《绵阳师范学院学报》
2016年第11期13-16,78,共5页
Journal of Mianyang Teachers' College
关键词
K-近邻法
分类
不等样本
有效性
Meanwhile, it can reduce the error ratios. Experiment shows the improvement is effective.