The subject of this textbook is deep learning, the modern incarnation of neural networks. This is the first textbook on this subject written by recognized academic authorities from within the field. The content of the book is organized into three parts: (1) mathematical preliminaries and introduction to relevant concepts in machine learning, (2) current deep learning techniques used by practitioners in industry, and (3) research perspectives for the future of deep learning. The scope is quite broad and encompasses nearly all major topics in deep learning with the exception of the recently developed field of deep reinforcement learning. This includes traditional neural networks as well as recurrent and convolutional networks, autoencoders, and deep probabilistic models. Forthcoming December 2016.