Machine Learning with TensorFlow.js

From Fundamentals to Production in JavaScript and TypeScript

Introduction: Why Machine Learning in JavaScript?

1. What This Book Covers
2. Who Should Read This Book
3. How to Use the Code Examples
4. A Note on Versions and Changes

Chapter 1: Machine Learning Meets JavaScript

1. Why Machine Learning in JavaScript?
2. How TensorFlow.js Fits Into the ML Landscape
3. Your First Prediction in 30 Seconds
4. Browser vs Node.js: Two Environments, One API
5. Setting Up Your Development Environment
6. Summary

Chapter 2: Tensors, the Building Blocks

1. What Is a Tensor?
2. Creating Tensors in TensorFlow.js
3. Tensor Shapes, Ranks, and Dimensions
4. Element-wise Operations and Broadcasting
5. Reshaping, Slicing, and Indexing
6. The Mathematics Behind the Operations
7. Summary

Chapter 3: Memory Management and Performance Basics

1. How TensorFlow.js Manages Memory
2. Tensors, GPU Buffers, and Garbage Collection
3. Disposing Tensors and Avoiding Leaks
4. tf.tidy and Automatic Cleanup
5. Monitoring Memory Usage
6. Performance Profiling Basics
7. Summary

Chapter 4: The Sequential API and Core Layers

1. Understanding Neural Networks Intuitively
2. Building Your First Model with Sequential
3. Dense Layers and Linear Transformations
4. Activation Functions and Non-linearity
5. Convolutional Layers for Spatial Data
6. Pooling and Dimensionality Reduction
7. Summary

Chapter 5: Training Models from Scratch

1. The Training Process Step by Step
2. Loss Functions: Measuring Model Quality
3. Optimizers and Learning Rate Strategies
4. Compiling a Model
5. The fit() Method and Training Loop
6. Callbacks for Monitoring and Control
7. Summary

Chapter 6: Evaluation, Prediction, and Debugging

1. Evaluating Model Performance
2. Making Predictions with model.predict()
3. Understanding Confusion Matrices and Metrics
4. Debugging Training Problems
5. Overfitting, Underfitting, and Regularization
6. Learning Curves and Diagnostics
7. Summary

Chapter 7: Advanced Model Architectures

1. The Functional API for Complex Graphs
2. Recurrent Layers and Sequence Modeling
3. LSTM and GRU Networks
4. Attention Mechanisms in TensorFlow.js
5. Custom Layers and Operations
6. Hyperparameter Tuning Strategies
7. Summary

Chapter 8: Data Handling and Preprocessing

1. The Data Pipeline in TensorFlow.js
2. Loading Data from Various Sources
3. Normalization and Standardization
4. One-hot Encoding and Label Transformation
5. Image Preprocessing and Augmentation
6. Text Tokenization and Embedding
7. Summary

Chapter 9: Computer Vision with TensorFlow.js

1. Image Classification Fundamentals
2. Using Pretrained Models: MobileNet and ResNet
3. Transfer Learning for Custom Classifiers
4. Object Detection with COCO-SSD
5. Body Pose Estimation and Face Detection
6. Building a Custom Vision Application
7. Summary

Chapter 10: Natural Language Processing in the Browser

1. Representing Text as Numbers
2. Sentiment Analysis with Pretrained Models
3. Building a Text Classifier from Scratch
4. Sequence-to-Sequence Models
5. Word Embeddings in TensorFlow.js
6. Practical NLP Applications
7. Summary

Chapter 11: Time-Series Forecasting and Sequential Data

1. Understanding Time-Series Data
2. Preparing Sequences for Modeling
3. Simple Baseline Models
4. LSTM-Based Forecasting
5. Evaluating Time-Series Predictions
6. Real-World Forecasting Example
7. Summary

Chapter 12: Saving, Loading, and Model Formats

1. The TensorFlow.js Model Format
2. Saving Models in the Browser
3. Saving Models in Node.js
4. Converting Python Keras Models to TensorFlow.js
5. Loading Models from Different Sources
6. Quantization and Model Compression
7. Summary

Chapter 13: Backend Systems and GPU Acceleration

1. How Backends Work in TensorFlow.js
2. The WebGL Backend and GPU Compute
3. WebGPU Backend and the Future
4. CPU Backend Fallbacks
5. TensorFlow C++ Backend for Node.js
6. Choosing and Configuring Backends
7. Summary

Chapter 14: Server-Side Machine Learning with Node.js

1. Running TensorFlow.js in Node.js
2. Building an Inference API
3. Batch Processing and Efficiency
4. Scaling Inference Services
5. Integration with Web Frameworks
6. Production Monitoring and Observability
7. Summary

Chapter 15: Deployment, Security, and Best Practices

1. Deployment Architectures and Patterns
2. Model Security and Protection
3. Privacy Considerations in Browser ML
4. Interoperability with Other ML Ecosystems
5. Testing Machine Learning Code
6. The Future of TensorFlow.js
7. Summary

Conclusion

Glossary

References