剪辑近邻法当不同类别的样本在分布上有交迭部分的，分类的错误率主要来自处于交迭

区的样本。当我们得到一个作为识别用的参考样本集时，由于不同类别交迭区域

中不同类别的样本彼此穿插，导致用近邻法分类出错。因此如果能将不同类别交

界处的样本以适当方式筛选，可以实现既减少样本数又提高正确识别率的双重目

的。为此可以利用现有样本集对其自身进行剪辑。

-Clips neighbor method when samples of different types of distribution in the overlap part of the classification error rate in the sample overlap mainly　the region. When we get a reference sample set identification purposes, since different types of samples in different types of overlapping regions interspersed with each other, resulting in classification error with the nearest neighbor. So if you can sample the different categories in an appropriate manner at the junction of screening, can not only reduce the number of samples and the dual purpose of improving the rate of correct identification. To do this you can use the existing sample set its own for editing.