This book is ideally suited to people who want a meaningful introduction into the most important contemporary concepts in Deep Learning. The book is accessible to people who lack both programming and linear algebra. Neither are needed to get a full understanding of everything the book offers.
IMO, the greatest moments in the book are the asides that appear in every chapter. The author will take a paragraph to note in passing things like '... no one really knows for sure why batch normalization helps. There are various hypotheses, but no certitudes." Or, "Importantly, I would generally recommend placing the previous layer's activation after the batch normalization layer (although this is still a subject of debate)." There is even an entire chapter dedicated to musings on the future of Deep Learning and general AI. This is the cherry on top that you don't get with most offers. Chollet offers them in nearly every chapter.
The book may as well have been called "Deep Learning with Keras" and that's not a bad thing. All the code is freely downloadable and can be run for free on a Google platform. You can freely ignore the implementation details and Python and simply run and learn from the notebooks provided. NOTE: As of February 2022, the new M1 Macs have bugs in the implementation of tensorflow that prevent a few code samples from working correctly. AND, some examples take so long to run (many hours) that there may be issues running them at Google. Frustrating though it might be, it does not detract from the experience.
As to cons, I don't see enough to warrant taking a star off the review. All important concepts are covered at an introductory level. The code works. The writing is clear. The author is an expert. There is a bizarre convention of having diagrams flow from the bottom to the top instead of top-down.
It's a good intro and basic reference. You'll get into more depth by taking the OpenAI courses at Coursera, but I'd actually recommend those as a next step after fully absorbing this book. Recommended.
While the book is titled "Deep Learning with Python", it might have been better titled, "Deep Learning with Keras." While Python is ostensibly
Enjoy fast, free delivery, exclusive deals, and award-winning movies & TV shows with Prime
Try Prime
and start saving today with fast, free delivery
Download the free Kindle app and start reading Kindle books instantly on your smartphone, tablet, or computer - no Kindle device required.
Read instantly on your browser with Kindle for Web.
Using your mobile phone camera - scan the code below and download the Kindle app.
 |
Frequently bought together
This item: Deep Learning with R, Second Edition
$53.20
Get it as soon as Monday, Aug 5
In Stock
+
$49.99
Get it as soon as Tuesday, Aug 6
In Stock
+
$66.00
Get it Aug 7 - 9
Only 5 left in stock - order soon.
Total price:
To see our price, add these items to your cart.
Choose items to buy together.
Customers who bought this item also bought
Page 1 of 1 Start overPage 1 of 1
About the author
Follow authors to get new release updates, plus improved recommendations.
Reviews with images
Excellent book and quality!!
Excellent book and quality!!! I love it super well explained
Top reviews from the United States
There was a problem filtering reviews right now. Please try again later.
Reviewed in the United States on February 3, 2022
Reviewed in the United States on December 27, 2023
Reviewed in the United States on August 31, 2023
Reviewed in the United States on May 27, 2022
Reviewed in the United States on October 11, 2023
Reviewed in the United States on August 31, 2023
Top reviews from other countries
SoulFireMage
Absolutely excellent in depth practical explanation
Reviewed in the United Kingdom on July 14, 2023
SoulFireMage
Reviewed in the United Kingdom on July 14, 2023
This one is well worth a second or third read if you are interested in Tensorflow and Keras or neural networks in general.
His crumpled paper analogy early on alerted me to the intuitive depth of this book. You do need just enough understand of linear algebra to appreciate what a vector space is to fully appreciate this analogy (this is not as hard as it sounds either, I promise. If you can do the first few videos on Khan Academy for linear algebra or half hour with a good tutorial, you'll be more than fine).
He avoids too much reliance on equations as the means of explanation, which for me, with barley enough linear algebra to tell a dot from a cross product, is great.
To be clear, it is best (in this whole field actually) if you understand differential calculus sufficient to appreciate the result of the power rule is a function not a number, and the afore mentioned linear algebra. If you spend an hour assuring yourself you can at least vaguely grasp this, then the more involved explanations will be very straightforwards. I emphasise this aspect because it's my own weakness - I have to work to focus on simple equations.
To make simple to intermediate models, following this book you still won't need much, if any depth of the maths skill, so don't let my mention of it out you off. It's a great solidly practical and wide ranging exploration of doing many tasks with deeplearning.
I recommend it with the Coursera developer certificate from DeepAi to any total beginner in neural networks.
Python knowledge needed?
I've coded with Python for approximately ten hours of actual experience but I am a software developer. So either good experience in other languages or the ability to work with simple python constructs and classes. A beginners course will help suffice it if you really don't have either.
The second book pictured is also excellent, worth taking in when you've gotten halfway through this one. The author recommends it and he's absolutely right.
Images in this review
