No. 345 of 500 No. 31 in Deep Learning

Deep Learning Illustrated

A Visual, Interactive Guide to Artificial Intelligence

By Jon Krohn, Grant Beyleveld and Aglaé Bassens

Published
September 18, 2019
Print length
416 pages
Reading level
Intermediate
ISBN-13
978-0135116692
Publisher
Addison-Wesley Professional
Written for
Python developers and analysts seeking a visual, code-based first course in deep learning.

Synopsis

Jon Krohn, Grant Beyleveld and illustrator Aglaé Bassens introduce deep learning by pairing color illustrations and analogies with runnable Python code in Jupyter notebooks. The book opens with four chapters that place the field in context through machine vision, natural language, machine-generated art and game-playing systems such as AlphaGo, before turning to essential theory: artificial neurons, dense networks, cost functions, gradient descent, backpropagation, and techniques for improving training such as dropout and careful weight initialization. Mathematics is kept to a minimum, with equations introduced only when needed to follow the code.

Later chapters apply that foundation to convolutional networks for vision, word vectors and recurrent networks for language, generative adversarial networks, and deep reinforcement learning, mostly using Keras on TensorFlow, with PyTorch introduced as an alternative. A closing chapter advises readers on starting their own projects and choosing libraries. Readers gain working intuition for how networks learn and hands-on experience building them, which makes later study of transformers and large language models easier. Because the book was published in 2019, it predates the transformer-dominated era, and reviewers have noted that its mathematical treatment is too light for readers who want to understand why the methods work.

Scorecard

Editorial score

How it scored

Editorial score: 68.0 / 1005 weighted criteria

  • Influence30%

    19.5/30
  • Authority25%

    17.5/25
  • Practical value20%

    15.0/20
  • Currency15%

    7.5/15
  • Readability10%

    8.5/10
  • Editorial score

    68.0/100

How the rank was computed

Rank No. 345 of 500 comes from a composite of 51.5: observable signals 20.8 of 55 points plus editorial review 30.7 of 45 points.

All 11 inputs
InputPoints
Observable signals20.8/55
Reach (Amazon ratings volume)8.6/16.5
Consensus across best-of lists5.5/13.8
Awards and bestseller evidence0.0/8.3
Reader rating, sample-adjusted5.9/8.3
Amazon sales rank0.8/5.5
Recency0.0/2.8
Editorial review30.7/45
Author authority7.9/11.3
Influence8.8/13.5
Practical value6.8/9.0
Readability3.8/4.5
Currency3.4/6.8

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 friendly, code-first tour of deep learning's major architectures, now dated on transformers but still effective for building intuition.

Strengths

  • Runnable notebooks accompany every major concept
  • Gentle on mathematics, strong on intuition
  • Broad coverage from vision to reinforcement learning

Keep in mind

Published in 2019, it omits transformers and large language models, and readers wanting mathematical depth will need a more rigorous companion text.

Key takeaways

  1. Neural networks learn by adjusting weights to minimize a cost function, using gradient descent with backpropagation to distribute error through the layers.
  2. Convolutional layers exploit spatial structure in images, which is why they transformed machine vision after AlexNet's 2012 ImageNet result.
  3. Word vectors give models numerical representations of meaning, a building block that later transformer-based language models extended.
  4. Generative adversarial networks pit a generator against a discriminator, while deep reinforcement learning trains agents through reward signals in simulated environments.

Notable facts

  • Published in seven languages, with editions including German, Korean, Russian, Traditional Chinese and Polish. Source (opens in a new tab)
  • Named a 2021 finalist in the Instructional, Technical, or Textbook category of the Data Community Content Creator Awards. Source (opens in a new tab)

About the author

Jon Krohn, Grant Beyleveld and Aglaé Bassens

Jon Krohn holds a doctorate in neuroscience from Oxford University, hosts the SuperDataScience podcast and cofounded Y Carrot. When the book appeared he was chief data scientist at the machine learning startup untapt, and he has guest lectured at Columbia University and NYU. Grant Beyleveld was a doctoral candidate at the Icahn School of Medicine at Mount Sinai; Aglaé Bassens is a Brooklyn-based Belgian artist.

Questions

Is Deep Learning Illustrated good for beginners?
It suits newcomers to deep learning who already know some Python. The illustrations, analogies and minimal mathematics make concepts approachable, and Jupyter notebooks let readers run every example. Complete programming novices will struggle with the code, and readers wanting rigorous derivations should follow it with a more mathematical textbook.
Does Deep Learning Illustrated cover transformers or ChatGPT-style models?
No. Published in 2019, it covers dense networks, convolutional networks, word vectors, recurrent networks, generative adversarial networks and deep reinforcement learning, but not the transformer architecture behind today's large language models. The fundamentals it teaches, including backpropagation, embeddings and training techniques, remain the foundation on which transformer-focused books build.
Which frameworks does Deep Learning Illustrated use?
Most examples use Keras running on TensorFlow, and the book also introduces PyTorch as the leading alternative library. All of the code is available open source on GitHub, and the authors produced about 18 hours of companion video tutorials that parallel the book's content, which helps readers who prefer to learn by watching code being written.

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.