On Learning Over-parameterized Neural Networks: A Functional Approximation Prospective

Open in new window