使用模糊加权距离的模糊K—近邻算法

FUZZY K-NN ALGORITHM USING FUZZY WEIGHTED DISTANCE

本文用自定义的模糊加权距离代替K—近邻分类器中的明氏距离,这种替代突出了每一样本中占有优势的特征分量对距离的贡献。仿真实验及实用结果表明这种替换可进一步改善分类器的性能。

ABSTRAST The Minkowski distance is replaced by the selfdefined fuzzy weiohbed-distance in k-nearest neighbor classifier. This replacement emphasizss the contribution to distance of the feature elements wmich have the superity in each sample, Simulation and practical results show that this replacement can further imp rov the property of classifier.