No. 192 of 500 No. 17 in Deep Learning

Dive into Deep Learning

By Aston Zhang, Zachary C. Lipton, Mu Li and Alexander J. Smola

Published
December 7, 2023
Print length
574 pages
Reading level
Intermediate
ISBN-13
978-1009389433
Publisher
Cambridge University Press
Written for
University students, instructors and engineers who want a rigorous, code-first deep learning textbook.

Synopsis

Dive into Deep Learning grew out of an open-source project, d2l.ai, drafted entirely in Jupyter notebooks that combine explanation, mathematics and executable code. The authors, scientists at Amazon Web Services along with Carnegie Mellon's Zachary Lipton, designed it so a reader can change a line of code and immediately see how a model behaves. The online version supports PyTorch, MXNet, JAX and TensorFlow, and Cambridge University Press publishes the first volume in print.

The printed volume moves from preliminaries such as data handling, linear algebra, calculus and probability through linear models for regression and classification, multilayer perceptrons, and a builders' guide to constructing custom layers and managing parameters. It then covers convolutional networks and modern architectures such as ResNet, recurrent networks including LSTMs and GRUs, and attention mechanisms and transformers. Later material on optimization, computational performance, computer vision, natural language processing, reinforcement learning, Gaussian processes and hyperparameter optimization continues online. The book assumes programming ability but no prior machine learning, and it supplies a mathematics refresher, yet the pace is that of a university course. Readers gain both the theory and the implementation habits needed to build models of their own.

Scorecard

Editorial score

How it scored

Editorial score: 80.3 / 1005 weighted criteria

  • Influence30%

    25.5/30
  • Authority25%

    21.3/25
  • Practical value20%

    16.0/20
  • Currency15%

    10.5/15
  • Readability10%

    7.0/10
  • Editorial score

    80.3/100

How the rank was computed

Rank No. 192 of 500 comes from a composite of 56.9: observable signals 20.7 of 55 points plus editorial review 36.2 of 45 points.

All 11 inputs
InputPoints
Observable signals20.7/55
Reach (Amazon ratings volume)7.7/16.5
Consensus across best-of lists5.5/13.8
Awards and bestseller evidence0.0/8.3
Reader rating, sample-adjusted4.8/8.3
Amazon sales rank1.0/5.5
Recency1.7/2.8
Editorial review36.2/45
Author authority9.6/11.3
Influence11.5/13.5
Practical value7.2/9.0
Readability3.2/4.5
Currency4.7/6.8

Each criterion is scored from 1 to 10 and weighted, and the points add up to the editorial score. The rank combines observable signals (55 percent) with this score (45 percent). Read the methodology.

Verdict

A rigorous, openly available textbook that teaches deep learning by pairing mathematics, prose and runnable code on nearly every page.

Strengths

  • Runnable code alongside every key concept
  • Supports PyTorch, JAX, TensorFlow and MXNet
  • Free online edition with discussion forums

Keep in mind

Its math density and course-level pace can overwhelm self-learners without calculus and linear algebra, and the print volume stops at transformers.

Key takeaways

  1. Deep learning is best learned by implementing models from scratch first, then reimplementing them concisely with a framework's high-level APIs.
  2. Casting a problem correctly, choosing the loss, model and data pipeline, matters as much as the network architecture itself.
  3. Modern convolutional designs such as ResNet show how architectural ideas like residual connections make very deep networks trainable.
  4. Attention mechanisms generalize earlier sequence models, and transformers built on them now underpin language, vision and multimodal systems.

Notable facts

About the author

Aston Zhang, Zachary C. Lipton, Mu Li and Alexander J. Smola

Aston Zhang and Mu Li were senior scientists at Amazon Web Services when they wrote the book with Alexander J. Smola, then an AWS vice president and distinguished scientist and formerly a Carnegie Mellon professor, and Zachary C. Lipton of Carnegie Mellon University. Smola and Li later cofounded Boson AI, where Smola is CEO and Li is CTO.

Questions

What is Dive into Deep Learning about?
It is a textbook that teaches deep learning through explanation, mathematics and runnable code side by side. The printed volume covers preliminaries, linear models, multilayer perceptrons, convolutional and recurrent networks, and attention and transformers. The free online edition continues with optimization, computer vision, natural language processing, reinforcement learning and more, with code in several frameworks.
Is Dive into Deep Learning free?
The full text is available at no cost on d2l.ai under a Creative Commons license, with downloadable notebooks for running the code. Cambridge University Press sells a printed edition of the first volume for readers who prefer a physical copy. The online version also includes the later chapters that the print volume omits.
Who should read Dive into Deep Learning?
It suits university students, instructors building a course, and engineers who can program in Python and are willing to work through calculus, linear algebra and probability. Self-learners without that math may find the pace demanding, though the appendix offers a refresher. Readers wanting a light conceptual overview of deep learning should choose a less technical book.

Sources

As an Amazon Associate, Best AI Books earns from qualifying purchases. Best AI Books is sponsored by Iternal.ai, whose founder and CEO, John Byron Hanby IV, wrote The AI Strategy Blueprint, our #1 ranked book. How we rank books.