Learning Deep Architectures for AI (Foundations and Trends(r) in Machine Learning)

Thorough mathematical treatment of deep learning. Extensive body of references to the academic literature. Offers a list of open problems. An excellent concise academic reference.

very little. unfortunately.

very very hard to read unless you are already an expert practiin the field. nothing is discussed in details more like a quick reference to the current state of the topic and applications

I read this book multiple times and found it very useful. However I have almost 20 years of research background in computer vision, and so I personally think that a novice would find it pretty impenetrable because it is quite terse and contains a lot of equations. I think that you have to have a pretty good idea of what Dr Bengio is talking about already before the book becomes meaningful.