How rotational invariance of common kernels prevents generalization
in high dimensions
Konstantin Donhauser 1 Mingqi Wu 1 Fanny Yang 1
Abstract conclusive in the regime d/nβ → c with β ∈ (0, 1). In par-
Kernel ridge regression is well-known to achieve ticular, even though several papers (Ghorbani et al., 2021;
minimax optimal rates in low-dimensional set- 2020; Liang et al., 2020a) show that the variance decreases ...


雷达卡




京公网安备 11010802022788号







