Practical Deep Learning for Cloud, Mobile, and Edge: Real-World AI & Computer-Vision Projects Using Python, Keras & TensorFlow 1st Edition

I am a deep learning practitioner and have been using TensorFlow building recommendation systems for production for several years. Learning the theory behind deep learning is one thing. Applying it to real-world scenarios is a whole another thing. What I really liked about this book is:1. The book stresses on building projects that work - first in as little lines of code as possible (and sometimes even without code), so the reader can get excited about something working. For example, the very initial autonomous car code (AWS DeepRacer) was 15 lines only. And then it dives into various aspects of how an engineer at an FAANG (or alike) company, would implement it at scale, benchmarking the cost, speed, memory, and energy usage of different approaches.2. It gave me an extensive overview of how deep learning is revolutionizing different industries (outside my own) through their projects and case studies.3. It puts the right words (and pictures) to deep learning theory and teaches how to apply them in real life using simple language. I came to learn deep learning through books full of math formulas and I am very surprised that this book managed to explain things in a very intuitive way without the heavy use of formulas.4. I found the “Creators Desk” boxes throughout the book were a nice touch, where people who built frameworks or products share their backstory behind how it came about.5. All the code has been made available on Github. Although I own a GPU, I can imagine how easy it is to train Keras and TensorFlow examples directly in one's browser without any framework installation (a rather tedious job) using Google Colab.

This is a really good book for getting into applied machine learning. It contains a lot of well thought out examples, corresponding code and case studies. If you are looking to dip your toes in the world of deep learning, this is a really good book for doing so.This book is going to be particularly helpful for the novice and enthusiast crowd. However, as mentioned in the preface, even experienced machine learning people can benefit by thinking about possible applications they would never have thought about.One of the best things about this book and relevant resources is that the corresponding code repository is actively being maintained and worked on. As with any software book, the book code may get outdated. My recommendation is that instead of relying on just the code in the book, actively look at the GitHub repository for the latest updates. The repo maintainers (authors, helpers, and likely altruistic readers) have done a good job of keeping the code up-to-date with newer changes to the interfaces.As a person coming in from a deeper ML background, one of my (minor) gripes, is that I find some citations lacking. However, unlike researchers, this book is more meant for people who are just beginning in this field, and thus it's not really a needed thing.Kudos for addressing the point of bias in the data, and the lack of explainability of the models. These are indeed a major problem in a lot of ML systems being deployed in production, and I believe that every person, ranging from the novice to the expert, needs to be aware of, and more importantly, make sure they mitigate it as much as possible.In terms of the implementation aspects, I really enjoyed the descriptions of the components and giving concrete examples of how to go about implementing the application in various platforms (cloud servers, android, IOS, javascript app, robots, take your pick.)To conclude, this is a really good book, with something for every level of ML practitioner, and even for people with no experience in ML.

I am a recent high school graduate, going into my freshmen year of college, who is interested in getting into the field of Deep Learning and AI. I found this book to be an incredible resource for accelerating my progress in these fields.Going in with absolutely no prior experience, I read this book and was astounded by how much it taught me in just a couple of days / weeks. The clear syntax, understandable language, and engaging examples support an excellent entrance into fields of Deep Learning, Computer Vision, and AI by not only providing easily accessible opportunities for hands-on learning, but also explaining the importance of the book's content. This was incredibly important to me as someone who did not know much about Deep Learning prior to reading this book: the way Koul, Ganju, and Kasam wrote this book goes beyond informative - they inspire creativity and practicality in the future of what Deep Learning, etc. can do.Further, Practical Deep Learning also provides an excellent transition from "I don't know anything" to "Wow! I am creating my own project... and understanding it!" Thus, the detail and content of this book make it suitable for all levels. It's a truly fantastic buy for anyone interested in Deep Learning - especially for those on the outside who are intimidated, like I was, by the concept of this field!