CAJ | 학술논문

文章提出一种融合互近邻和可信度的 K近邻算法，根据互近邻的概念删除噪声数据；利用由近邻诱导待分类样本标签的可信度，避免待分类样本近邻中大类吃小类的概率。该算法不仅可以减小噪声数据对分类的影响，而且一定程度上增强了 K近邻分类算法的稳定性。该算法在UCI标准数据集上进行了测试，性能相当或优于其他分类器。
문장제출일충융합호근린화가신도적 K근린산법，근거호근린적개념산제조성수거；이용유근린유도대분류양본표첨적가신도，피면대분류양본근린중대류흘소류적개솔。해산법불부가이감소조성수거대분류적영향，이차일정정도상증강료 K근린분류산법적은정성。해산법재UCI표준수거집상진행료측시，성능상당혹우우기타분류기。
The algorithm for K-nearest neighbor(KNN) classification which combines mutual neighbors and credibility is presented .The noise data is eliminated according to the concept of mutual neighbor . Then the credibility of neighbor-induced labeling of the sample to be classified is used to avoid the probability of the merging of subcategories by categories .T he algorithm can not only reduce the im-pact of noise on the classification of data ,but also to some extent ,enhance the stability of KNN clas-sification algorithm .The presented method is tested on the UCI datasets ,and the results show that the proposed technique is better than or equal to other classifiers .