Matrix calculus for deep learning

This paper is an attempt to explain all the matrix calculus you need in order to understand the training of deep neural networks.