Preprint
Jul 2026
Upper Confidence Bounds for the Prediction Error of Kernel Ridge Regression via Gaussian Refitting
This work proposes a Gaussian refit for kernel ridge regression by Anderson's inequality, which requires no moment assumptions and is calibrated at any confidence level via order statistics, and extends empirically to nonlinear constrained estimators and real spatial data.
Yijin Ni, Xiaoming Huo
· 2 citations