Hardness of Learning Neural Networks under the Manifold Hypothesis

Open in new window