摘要
支持向量机是近年来发展的以统计学习理论为基础的新型学习机。该学习机用结构风险代替经验风险,因而具有传统的神经网络无法相比的优势。在该学习机的各各研究方向中,核函数的选择无疑是极其重要的核心问题。通过对核矩阵的计算和研究,从理论上为核函数的选择提供了参考。
Support Vector Machines were developed in recent years,which base on statistical learning theory.They replace the experiential risk by structural risk,thus have a large advantage over the traditional neural network.In all research fields of the learning machines,the selection of kernel function is the most important problem.A reference for selecting the kernel function for SVM theoretically is given,through observing and computing the kernel matrix.
出处
《科学技术与工程》
2008年第16期4513-4517,共5页
Science Technology and Engineering
关键词
支持向量机
核函数
模型选择
结构风险
核矩阵
support vector machine kernel function model selection structural risk kernel matrix
