Quantifying the generalization error in deep learning in terms of data distribution and neural network smoothness

Open in new window