论文标题
对于插值内核机，将ERM溶液的标准最小化可最大程度地减少稳定性
For interpolating kernel machines, minimizing the norm of the ERM solution minimizes stability
论文作者
论文摘要
我们研究平均$ \ mbox {cv} _ {loo} $无脊回归的稳定性，并得出相应的风险范围。我们表明，具有最小规范的插值解决方案最小化了$ \ mbox {cv} _ {loo} $稳定性上的界限，这又由经验内核矩阵的条件编号控制。后者可以在渐近状态下进行表征，在渐近方案中，数据的维度和基数都归功于无穷大。在随机核矩阵的假设下，应期望相应的测试误差遵循双下降曲线。
We study the average $\mbox{CV}_{loo}$ stability of kernel ridge-less regression and derive corresponding risk bounds. We show that the interpolating solution with minimum norm minimizes a bound on $\mbox{CV}_{loo}$ stability, which in turn is controlled by the condition number of the empirical kernel matrix. The latter can be characterized in the asymptotic regime where both the dimension and cardinality of the data go to infinity. Under the assumption of random kernel matrices, the corresponding test error should be expected to follow a double descent curve.
