I recently finished Part I of Deep Learning by Goodfellow, Bengio, and Courville. It was a clear and efficient primer on the mathematical preliminaries needed to engage more deeply with the subject. I appreciated the refresher in linear algebra, probability, and numerical computation, which helped reestablish important foundations. In places, I found myself wishing for more proofs—or at least proof sketches—as I often wanted to understand where results came from rather than accept them at face value.
The book is refreshingly direct and avoids unnecessary digressions. Chapter 5, in particular, served as a solid recap of core machine learning ideas. Overall, this section left me motivated to continue further into the theoretical side of deep learning, complementing the more application-focused perspective I’ve gained elsewhere.