论文标题
随机设计对山脊回归的基本分析
An elementary analysis of ridge regression with random design
论文作者
论文摘要
在本说明中，我们通过随机设计提供了脊回归的预测误差的基础分析。证明是简短而独立的。特别是，它通过交换性参数，矩阵扰动和操作员凸性的结合，绕过了鲁德森对协方差矩阵的偏差不平等的使用。
In this note, we provide an elementary analysis of the prediction error of ridge regression with random design. The proof is short and self-contained. In particular, it bypasses the use of Rudelson's deviation inequality for covariance matrices, through a combination of exchangeability arguments, matrix perturbation and operator convexity.
