On the Generalization Power of Overfitted Two-Layer Neural Tangent Kernel Models

MLT
Re-assign community