New sections on autoencoders and the word2vec network within the multilayer perceptrons chapter.

This edition features substantial updates to reflect the rapid evolution of the field since the previous release:

New material on deep reinforcement learning, policy gradient methods, and the use of deep networks within the RL framework.