改良最短距离法在合金样品分类中的应用研究

Application of Alternative k-Nearest Neighbour to Classification of Alloy Samples

摘要: 运用改良最短距离法分类原理对合金样品进行分类研究，从样本个数，训练集特征值范围及试验集插入位置等方面对识别率的影响进行了探讨。改良最短距离法对未知样品有较高的识别率，但该方法的外推预测能力不足。

Abstract: The alternative knearest neighbour (ALKNN) method is briefly introduced in this paper and study on its application to classification of alloy samples is reported. The software was designed by using SDK and C language. The effect of number of samples, the