Just in case you didn’t know, Adrew Ng is writing a book.
He announced it in NIPS 2016, and has been sending chapters over his mailing list. The book is called ‘Machine Learning Yearning’. How is it different from the others? It’s chock full of real life, actionable wisdom. Each chapter is 2-3 pages. It’s dense. It uses different naming for training/test/holdout (training/dev/test)… but it’s really effective. If you are solving problems with deep learning at work today, I’m sure you are facing the problems he writes about. It’s the closest thing to a typical O’reilly cookbook for machine learning.
Sorry I’m not posting a direct link; he’s sending chapters over email as he finishes them. I’m not affiliated with Andrew in any way, just sharing the resource as I found it very useful.