论文部分内容阅读
依据 KMSE 模型对应的特征空间中的鉴别矢量可表示为部分训练样本的线性组合这一理论前提,可利用回归分析中变量选择的思路对 KMSE 模型加以改进.在本文中为了提高 KMSE 的分类效率而发展出的基于最小平方误差准则的算法能大大提升 KMSE 模型的分类速度.实验结果显示该算法还能取得较优的分类性能.
According to the theoretical premise of linear combination of some training samples, the KMSE model can be improved by using the idea of variable selection in regression analysis.In this paper, in order to improve the classification efficiency of KMSE The developed algorithm based on least square error criterion can greatly improve the classification speed of KMSE model.The experimental results show that the algorithm can achieve better classification performance.