|
|||||
|
|
| 基于SVM的特征加权KNN算法 | |
| 引用本文： | 陈振洲,李磊,姚正安.基于SVM的特征加权KNN算法[J].中山大学学报(自然科学版),2005,44(1):17-20. |
| 作者姓名： | 陈振洲 李磊 姚正安 |
| 作者单位： | 1. 中山大学软件研究所,广东,广州,510275;中山大学数学与计算科学学院,广东,广州,510275 2. 中山大学软件研究所,广东,广州,510275 3. 中山大学数学与计算科学学院,广东,广州,510275 |
| 摘 要： | 作为一种非参数的分类算法,K-近邻(KNN)算法是非常有效和容易实现的.它已经广泛应用于分类、回归和模式识别等.在应用KNN算法解决问题的时候,要注意两个方面的问题--样本权重和特征权重.利用SVM来确定特征的权重,提出了基于SVM的特征加权算法(FWKNN,feature weighted KNN).实验表明,在一定的条件下,FWKNN能够极大地提高分类准确率. |
| 关 键 词： | 支持向量机 K-近邻算法 距离加权 特征加权 |
| 文章编号： | 0529-6579(2005)01-0017-04 |
| 修稿时间： | 2004年3月5日 |
Feature-Weighted K-Nearest Neighbor Algorithm with SVM |
|
| CHEN Zhen_zhou.Feature-Weighted K-Nearest Neighbor Algorithm with SVM[J].Acta Scientiarum Naturalium Universitatis Sunyatseni,2005,44(1):17-20. | |
| Authors: | CHEN Zhen_zhou |
| Institution: | CHEN Zhen_zhou~ |
| Abstract: | As a nonparametric approach, K_Nearest Neighbor (K_NN) algorithm is very efficient and easily to be realized. The k_NN algorithm has been successfully used in classification, regression and pattern recognition. Two aspects, the weight of samples and the weight of features, must be paid attention to when applying K_NN to solve problems. SVM (support vector machine) to quantify the weight of features was used and FWKNN (feature_weighted K_nearest neighbor algorithm with SVM) was proposed. Experiments on artificial and natural datasets FWKNN was proposed showed that, in most cases, FWKNN improve the accuracy of classification. |
| Keywords: | SVM K_nearest neighbor distance_weighted feature_weighted |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ | |