Practical AI on the Google Cloud Platform

Table of contents

1. Preface
   1. Who Should Read This Book
   2. Why I Wrote This Book
   3. Navigating This Book
   4. A Note on the Google AI Platform
   5. Things You Need for This Book
   6. Conventions Used in This Book
   7. Using Code Examples
   8. O’Reilly Online Learning
   9. How to Contact Us
   10. Acknowledgments
2. 1. Data Science and Deep Learning
   1. What Is Data Science?
   2. Classification and Regression
      1. Regression
      2. Goodness of Fit
      3. Classification with Logistic Regression
      4. Multivariant Regression and Classification
   3. Data Discovery and Preparation
      1. Bad Data
      2. Training, Test, and Validation Data
      3. Good Data
      4. Preparing Data
      5. Questioning Your Data
   4. The Basics of Deep Learning
      1. The Perceptron Game
   5. Understanding How Networks Learn
      1. Backpropagation
      2. Optimization and Gradient Descent
      3. Vanishing or Exploding Gradients
      4. SGD and Batching Samples
      5. Batch Normalization and Regularization
      6. Activation Functions
      7. Loss Functions
   6. Building a Deep Learner
      1. Optimizing a Deep Learning Network
      2. Overfitting and Underfitting
      3. Network Capacity
   7. Conclusion
      1. Game Answers
3. 2. AI on the Google Cloud Platform
   1. AI Services on GCP
      1. The AI Hub
      2. AI Platform
      3. AI Building Blocks
   2. Google Colab Notebooks
      1. Building a Regression Model with Colab
   3. AutoML Tables
   4. The Cloud Shell
   5. Managing Cloud Data
   6. Conclusion
4. 3. Image Analysis and Recognition on the Cloud
   1. Deep Learning with Images
      1. Enter Convolution Neural Networks
   2. Image Classification
      1. Set Up and Load Data
      2. Inspecting Image Data
      3. Channels and CNN
      4. Building the Model
      5. Training the AI Fashionista to Discern Fashions
      6. Improving Fashionista AI 2.0
   3. Transfer Learning Images
      1. Identifying Cats or Dogs
      2. Transfer Learning a Keras Application Model
      3. Training Transfer Learning
      4. Retraining a Better Base Model
   4. Object Detection and the Object Detection Hub API
      1. YOLO for Object Detection
   5. Generating Images with GANs
   6. Conclusion
5. 4. Understanding Language on the Cloud
   1. Natural Language Processing, with Embeddings
      1. Understanding One-Hot Encoding
      2. Vocabulary and Bag-of-Words
      3. Word Embeddings
      4. Understanding and Visualizing Embeddings
   2. Recurrent Networks for NLP
      1. Recurrent Networks for Memory
      2. Classifying Movie Reviews
      3. RNN Variations
   3. Neural Translation and the Translation API
      1. Sequence-to-Sequence Learning
      2. Translation API
      3. AutoML Translation
   4. Natural Language API
   5. BERT: Bidirectional Encoder Representations from Transformers
      1. Semantic Analysis with BERT
      2. Document Matching with BERT
      3. BERT for General Text Analysis
   6. Conclusion
6. 5. Chatbots and Conversational AI
   1. Building Chatbots with Python
   2. Developing Goal-Oriented Chatbots with Dialogflow
   3. Building Text Transformers
      1. Loading and Preparing Data
      2. Understanding Attention
      3. Masking and the Transformer
      4. Encoding and Decoding the Sequence
   4. Training Conversational Chatbots
      1. Compiling and Training the Model
      2. Evaluation and Prediction
   5. Using Transformer for Conversational Chatbots
   6. Conclusion
7. 6. Video Analysis on the Cloud
   1. Downloading Video with Python
   2. Video AI and Video Indexing
   3. Building a Webcam Face Detector
      1. Understanding Face Embeddings
   4. Recognizing Actions with TF Hub
   5. Exploring the Video Intelligence API
   6. Conclusion
8. 7. Generators in the Cloud
   1. Unsupervised Learning with Autoencoders
      1. Mapping the Latent Space with VAE
   2. Generative Adversarial Network
   3. Exploring the World of Generators
      1. A Path for Exploring GANs
      2. Translating Images with Pix2Pix and CycleGAN
   4. Attention and the Self-Attention GAN
      1. Understanding Self-Attention
      2. Self-Attention for Image Colorization—DeOldify
   5. Conclusion
9. 8. Building AI Assistants in the Cloud
   1. Needing Smarter Agents
   2. Introducing Reinforcement Learning
      1. Multiarm Bandits and a Single State
      2. Adding Quality and Q Learning
      3. Exploration Versus Exploitation
      4. Understanding Temporal Difference Learning
   3. Building an Example Agent with Expected SARSA
      1. Using SARSA to Drive a Taxi
      2. Learning State Hierarchies with Hierarchical Reinforcement Learning
   4. Bringing Deep to Reinforcement Learning
      1. Deep Q Learning
      2. Optimizing Policy with Policy Gradient Methods
   5. Conclusion
10. 9. Putting AI Assistants to Work
    1. Designing an Eat/No Eat AI
    2. Selecting and Preparing Data for the AI
    3. Training the Nutritionist Model
    4. Optimizing Deep Reinforcement Learning
    5. Building the Eat/No Eat Agent
    6. Testing the AI Agent
    7. Commercializing the AI Agent
       1. Identifying App/AI Issues
       2. Involving Users and Progressing Development
    8. Future Considerations
    9. Conclusion
11. 10. Commercializing AI
    1. The Ethics of Commercializing AI
    2. Packaging Up the Eat/No Eat App
    3. Reviewing Options for Deployment
       1. Deploying to GitHub
       2. Deploying with Google Cloud Deploy
    4. Exploring the Future of Practical AI
    5. Conclusion
12. Index