A comparison of deep networks with ReLU activation function and linear spline-type methods

Open in new window