Deep Learning from Scratch: Building with Python from First Principles

by Seth Weidman

With the resurgence of neural networks in the 2010s, deep learning has become essential for machine learning practitioners and even many software engineers. This book provides a comprehensive introduction for data scientists and software engineers with machine learning experience. Youâ??ll start with deep learning basics and move quickly to the details of important advanced architectures, implementing everything from scratch along the way. Author Seth Weidman shows you how neural networks work using a first principles approach. Youâ??ll learn how to apply multilayer neural networks, convolutional neural networks, and recurrent neural networks from the ground up. With a thorough understanding of how neural networks work mathematically, computationally, and conceptually, youâ??ll be set up for success on all future deep learning projects. This book

Genres: Artificial Intelligence, Nonfiction, Computer Science, Programming, Computers, Technical

250 pages, Paperback

Published October 15, 2019

Reviews

[Maria-Anna · Rating 4 out of 5]

The book does a great job in explaining complex things in more or less simple terms and with a bunch of visuals. Although it still feels like in some chapters the author was in a rush and skipped some of the details which could be discussed more. Probably the only real disadvantage of this book is the only one version of implementing things in code which is presented by the author. It would be nicer to get more alternatives of the possible implementations of described concepts and to hear from the author how to improve the solution he has shown.

[LeoQuiroa · Rating 5 out of 5]

If someone asks me what is the first book that I should read to start in the world of ML, I would definitely recommend this one. The level of implementation is so granular, especially in chapter#3 that you will feel like you finally grasp the whole concept and have that awesome aha-moment.

[Vỹ Hồng]

I really like how the author tries to explain each concept in 3 models: mathematical formula, box diagram, and code. It's quite refreshing and significantly aids understandings. The explanation is also quite clear and engaging.

However, there are many symbolic errors scattered throughout the book. Symbols used in diagrams often do not agree with the mathematical formula, and variable names in code. I see these errors once every few pages, and it's difficult to trust the content when I have to constantly substituting values. This must be one of the worst edited O'Reilly book I have ever read. I had to drop after a few chapters due to this issue.

[Matterhorn · Rating 3 out of 5]

The math notation is often not very formal and not consistent. It is good intention to link math, code and big picture together, but the lack of formalism with the math, makes it very difficult to read. The derivatives are all over the place.

[Jin Shusong · Rating 5 out of 5]

I like this book. It explain ML from basic concepts and algorithms.

[Thinh Nguyen · Rating 5 out of 5]

It is a thorough book for those who want to learn deep learning from its inner workings. Although the book does contain some minor errors (typo + code), its contents are well-written and well-organized. Concepts are also made simple and understandable for beginners. Still recommended!

[Chen Mingshuo · Rating 3 out of 5]

ugly math