The Connection Between Approximation, Depth Separation and Learnability in Neural Networks

Open in new window