摘要
本文将机器学习中K近邻(KNN)方法应用于动态心电图波形分类模型的建立.但KNN算法属于线性分类,因此引入了核函数的概念,将波形数据间的线性差异转化为非线性.本文主要对常见的核函数中的高斯核函数,四次样条核函数及改进的复合四次样条核函数进行了比较,实验结果显示,3种方法在一定程度上都提高了分类准确度.
In this paper, we use K nearest neighbor （KNN） method for hoher waveform classification. As KNN is a kind of linear classification model, we introduce kernel function method to KNN, in order to change the linear difference into nonlinear difference. In this paper, we compared Gaussian kernel and Spline kernel. The experiment result shows both of the two methods improve the accuracy of classification.
出处
《天津理工大学学报》
2009年第5期42-45,共4页
Journal of Tianjin University of Technology
基金
天津市高等高校科技发展基金(20061009)
