Generalization Guarantees for Neural Networks via Harnessing the Low-rank Structure of the Jacobian

Open in new window