Bernstein-Schur Kernels: Random Features by Sketched Modulation and Radial Randomization

In three linesBernstein-Schur kernels: random feature construction combining sketched finite modulation and radial randomization via Bernstein-Widder scale. Feature dimension Dm without O(d²) cost of exact modulation. Exact variance guarantees and operator-norm bounds controlled by intrinsic dimension, with kernel ridge regression applications.

Summary generated by Claude — human-verified