Ty
Bob Coecke's book assumes little knowledge. He and coauthor Aleks KISSINGER are clear writers.
The book uses a category theory based notation much like the one for angular momentum tensors in Roger PENROSE and Wolfgang RINDLER (*Spinors and spacetime, 1,2, Oxford: University Press, 1984, 1986). The notation does much of the work equations would do and, while a pain to typeset, makes it easy and a pleasure to reason about complex quantum systems and quickly conveys the gists of things.
Stuart RUSSELL and Peter NORVIG (*Artificial intelligence a modern approach, Upper Saddle River: Prentice Hall, 2010) is very comprehensive. It's written in the style of one of Donald Knuth's books and is very readable. The one issue is that it's technical yet not very deep; but it covers most of the definitions.
It's around 1000 pages, much like Margaret BODEN (*Mind as machine, Oxford: University Press, 2006).
My suggestion is to read RUS10 and BOD06 at the same time.
BOD06 is very entertaining and goes deeper than RUS10 into open problems, and this despite RUS10 having more mathematics.
Neither of them discuss the programming side of things.
All three books are self contained and can be read by someone new to the field but BOD06 and COE17 have much to recommend them to those already in the field.
If you want to program neural networks with convolution or reservoir layers and don't yet know much about programming, you may want to read Richard HAMMING (Digital filters, Engelwood Cliffs: Prentice Hall, 1977) and Bruce MACLENNON (Principles of programming languages, New York: Oxford University Press).
Thank you for your comprehensive reply!