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

Open in new window