Deep Learning 22: Variational Autoencoder : Derivation Of The Loss Function