O'Reilly logo

Stay ahead with the world's most comprehensive technology and business learning platform.

With Safari, you learn the way you learn best. Get unlimited access to videos, live online training, learning paths, books, tutorials, and more.

Start Free Trial

No credit card required

Deep Learning Illustrated: A Visual, Interactive Guide to Artificial Intelligence

Book Description

Deep learning is one of today's hottest fields. This approach to machine learning is achieving breakthrough results in some of today's highest profile applications, in organizations ranging from Google to Tesla, Facebook to Apple. Thousands of technical professionals and students want to start leveraging its power, but previous books on deep learning have often been non-intuitive, inaccessible, and dry. In Deep Learning Illustrated, three world-class instructors and practitioners present a uniquely visual, intuitive, and accessible high-level introduction to the techniques and applications of deep learning. Packed with vibrant, full-color illustrations, it abstracts away much of the complexity of building deep learning models, making the field more fun to learn and accessible to a far wider audience.


Part I's high-level overview explains what Deep Learning is, why it has become so ubiquitous, and how it relates to concepts and terminology such as Artificial Intelligence, Machine Learning, Artificial Neural Networks, and Reinforcement Learning. These opening chapters are replete with vivid illustrations, easy-to-grasp analogies, and character-focused narratives.


Building on this foundation, the authors then offer a practical reference and tutorial for applying a wide spectrum of proven deep learning techniques. Essential theory is covered with as little mathematics as possible and is illuminated with hands-on Python code. Theory is supported with practical "run-throughs" available in accompanying Jupyter notebooks, delivering a pragmatic understanding of all major deep learning approaches and their applications: machine vision, natural language processing, image generation, and videogaming.


To help readers accomplish more in less time, the authors feature several of today's most widely used and innovative deep learning libraries, including TensorFlow and its high-level API, Keras; PyTorch; and the recently released, high-level Coach, a TensorFlow API that abstracts away the complexity typically associated with building Deep Reinforcement Learning algorithms.

Table of Contents

  1. Cover Page
  2. Title Page
  3. Contents
  4. Table of Contents
  5. About the Authors
  6. Introduction
    1. How To Read This Book
    2. Acknowledgments
  7. Part I: Introducing Deep Learning
    1. 1 Biological and Machine Vision
      1. Biological Vision
      2. Machine Vision
      3. TensorFlow Playground
      4. Quick, Draw!
      5. Summary
    2. 2 Human and Machine Language
      1. Deep Learning for Natural Language Processing
      2. Computational Representations of Language
      3. Elements of Natural Human Language
      4. Google Duplex
      5. Summary
    3. 3 Machine Art
      1. A Boozy All-Nighter
      2. Arithmetic on Fake Human Faces
      3. Style Transfer: Converting Photos into Monet (and Vice Versa)
      4. Make Your Own Sketches Photorealistic
      5. Creating Photorealistic Images from Text
      6. Image Processing Using Deep Learning
      7. Summary
    4. 4 Game-Playing Machines
      1. Deep Learning, AI, and Other Beasts
      2. Three Categories of Machine Learning Problems
      3. Deep Reinforcement Learning
      4. Video Games
      5. Board Games
      6. Manipulation of Objects
      7. Popular Deep Reinforcement Learning Environments
      8. Three Categories of AI
      9. Summary
  8. Part II: Essential Theory Illustrated
    1. 5 The (Code) Cart Ahead of the (Theory) Horse
      1. Prerequisites
      2. Installation
      3. A Shallow Network in Keras
      4. Summary
    2. 6 Artificial Neurons Detecting Hot Dogs
      1. Biological Neuroanatomy 101
      2. The Perceptron
      3. Modern Neurons and Activation Functions
      4. Choosing A Neuron
      5. Summary
      6. Key Concepts
    3. 7 Artificial Neural Networks
      1. The Input Layer
      2. Dense Layers
      3. A Hot Dog-Detecting Dense Network
      4. The Softmax Layer of a Fast Food-Classifying Net-work
      5. Revisiting our Shallow Network
      6. Summary
      7. Key Concepts
    4. 8 Training Deep Networks
      1. Cost Functions
      2. Optimization: Learning to Minimize Cost
      3. Backpropagation
      4. Tuning Hidden-Layer Count and Neuron Count
      5. An Intermediate Net in Keras
      6. Summary
      7. Key Concepts
    5. 9 Improving Deep Networks
      1. Weight Initialization
      2. Unstable Gradients
      3. Model Generalization (Avoiding Overfitting)
      4. Fancy Optimizers
      5. A Deep Neural Network in Keras
      6. Regression
      7. TensorBoard
      8. Summary
      9. Key Concepts
  9. Part III: Interactive Applications of Deep Learning
    1. 10 Machine Vision
      1. Convolutional Neural Networks
      2. Pooling Layers
      3. LeNet-5 in Keras
      4. AlexNet and VGGNet in Keras
      5. Residual Networks
      6. Applications of Machine Vision
      7. Summary
      8. Key Concepts
    2. 11 Natural Language Processing
      1. Preprocessing Natural Language Data
      2. Creating Word Embeddings with word2vec
      3. The Area Under the ROC Curve
      4. Natural Language Classification with Familiar Networks
      5. Networks Designed for Sequential Data
      6. Non-Sequential Architectures: The Keras Functional API
      7. Summary
      8. Key Concepts
    3. 12 Generative Adversarial Networks
      1. Essential GAN Theory
      2. The “Quick, Draw!” Dataset
      3. The Discriminator Network
      4. The Generator Network
      5. The Adversarial Network
      6. GAN Training
      7. Summary
      8. Key Concepts
    4. 13 Deep Reinforcement Learning
      1. Essential Theory of Reinforcement Learning
      2. Essential Theory of Deep Q-Learning Networks
      3. Defining a DQN Agent
      4. Interacting with an OpenAI Gym Environment
      5. Hyperparameter Optimization with SLM Lab
      6. Agents Beyond DQN
      7. Summary
      8. Key Concepts
  10. Part IV: You and A.I.
    1. 14 Moving Forward with Your Own Deep Learning Projects
      1. Ideas for Deep Learning Projects
      2. Resources for Further Projects
      3. The Modeling Process, including Hyperparameter Tuning
      4. Deep Learning Libraries
      5. Software 2.0
      6. Approaching Artificial General Intelligence
      7. Summary
  11. Part V: Appendices
    1. Appendix A. Formal Neural Network Notation
    2. Appendix B. Backpropagatior
    3. Appendix C. PyTorch
      1. PyTorch Features
      2. PyTorch in Practice