https://yellokat.github.io/posts/7.-Regularization-for-Deep-Learning/