论文部分内容阅读
为了提高特征的分类性能,提出一种基于 K 近邻的决策边界分析(KNN-DBA)算法.该算法的决策边界由K近邻分类器决定,提取的特征维数不受类别数的限制,算法简单且速度快.在手写数字样本集 USPS 和 UCI 中的PenDigits 上用最近邻分类器和支持向量机(SVM)对决策边界分析进行验证,实验结果表明 KNN-DBA 识别性能优于主成分分析和基于 SVM 的决策边界分析.
In order to improve the performance of feature classification, a K-nearest neighbor decision boundary analysis (KNN-DBA) algorithm is proposed. The decision boundary of the algorithm is decided by K nearest neighbor classifier. The extracted feature dimension is not restricted by the number of classes, And the speed is fast.Experimental results show that the performance of KNN-DBA is better than that of principal component analysis and based on Principal Component Analysis (PCA) based on nearest neighbor classifier and Support Vector Machine (SVM) on PenDigits in USPS and UCI. Decision Boundary Analysis of SVM.