Hands-On Machine Learning with TensorFlow.js: A guide to building ML applications integrated with web technology using the TensorFlow.js library

Cover
Title Page
Copyright and Credits
About Packt
Contributors
Table of Contents
Preface
Section 1: The Rationale of Machine Learning and the Usage of TensorFlow.js
Chapter 1: Machine Learning for the Web
	Technical requirements
	Why machine learning on the web?
	Operation graphs
		Visualizing an operation graph
		Automatic differentiation
	What is TensorFlow.js?
		Seamless integration with web technologies
		Hardware acceleration using web browser APIs
		Compatibility with TensorFlow
		Data privacy
	Installing TensorFlow.js
		tfjs-converter
	The low-level API
		Tensors
		Operations
		Memory
		Eager execution
	The Layers API
		Sequential model
		Functional model
		Model summary
		Custom layers
	Summary
	Questions
	Further reading
Chapter 2: Importing Pretrained Models into TensorFlow.js
	Technical requirements
	The portable model format
		Protocol buffers
		GraphDef
		NodeDef
	Exporting a model from TensorFlow
		TensorFlow's SavedModel
			The SavedModel format
			Simple save
			The SavedModelBuilder API
		The Keras HDF5 model
	Converting models using tfjs-converter
		Converting a TensorFlow SavedModel
		The Keras HDF5 model
		The TensorFlow Hub module
	Loading the model into TensorFlow.js
		Supported operations
	Summary
	Questions
	Further reading
Chapter 3: TensorFlow.js Ecosystem
	Technical requirements
	Why high-level libraries?
	Using existing models
		MobileNet in tfjs-models
		Supported models
		Image classification application
		Example applications in the community
	Loading the data from various kinds of storage
		Data sources
		Webcam
	Pose detection with ML5.js
		Supported models
		PoseNet in ML5.js
	Drawing cats with Magenta.js
		Sketch drawing
	XOR classification with machinelearn.js
		Random forest classifier
	Summary
	Exercises
	Further reading
Section 2: Real-World Applications of TensorFlow.js
Chapter 4: Polynomial Regression
	Technical requirements
	What is polynomial regression?
		Supervised learning
		The simplest linear model
		General polynomial model
		The loss function
		Optimizer for machine learning
		Optimizers in TensorFlow.js
	Two-dimensional curve fitting
		Preparing the dataset
		Applying the 2-degree polynomial model
		Loss function by mean squared error
		Looking into the optimization process
		Fitting the curve
	Summary
	Questions
	Further reading
Chapter 5: Classification with Logistic Regression
	Technical requirements
	Background of binary classification
	What is logistic regression?
		The behavior of the probabilistic generative model
		Optimization process
	Classifying two-dimensional clusters
		Preparing the dataset
		Logistic regression model with the Core API
		Optimizing with the cross-entropy loss function
		Implementing a logistic regression model with the Layers API
		Implementing a logistic regression model with machinelearn.js
	Summary
	Questions
	Further reading
Chapter 6: Unsupervised Learning
	Technical requirements
	What is unsupervised learning?
	Learning how K-means works
		Centroid
		Algorithm
		Evaluation
	Generalizing K-means with the EM algorithm
		The algorithm
		Relationship with K-means
	Clustering two groups in a 2D space
		The three clusters
		K-means implementation
	Summary
	Exercise
	Further reading
Chapter 7: Sequential Data Analysis
	Technical requirements
	What is Fourier transformation?
		Constituent frequencies as a feature
		Discrete Fourier transform
		Fast Fourier transform
	Cosine curve decomposition
		Complex number type
		The cosine curve
		Fourier transformation for cosine curves
		Fourier transformation for composite curves
		Inversed Fourier transform
	Summary
	Exercise
	Further reading
Chapter 8: Dimensionality Reduction
	Technical requirements
	Why dimensionality reduction?
		Curse of dimensionality
	Understanding principal component analysis
		Variance maximization
	Projecting 3D points into a 2D space with PCA
		Three-dimensional clusters
		Principal component calculation
		The variance of projected datasets
	Word embedding
		What is word embedding?
		Loading the IMDb dataset
		Embedding the model
		Visualization of embeddings
	Summary
	Exercise
	Further reading
Chapter 9: Solving the Markov Decision Process
	Technical requirements
	Reinforcement learning
		MDP
		Discounted total reward
		State-value function
		Bellman equation
		Q-learning
	Solving the four-states environment
		Designing the environment
		The Q-learning process
	Summary
	Exercise
	Further reading
Section 3: Productionizing Machine Learning Applications with TensorFlow.js
Chapter 10: Deploying Machine Learning Applications
	Technical requirements
	The ecosystem around the JavaScript platform
		JavaScript in modern web browsers
		Node.js
		Node package manager
		Benefits of TypeScript in ML applications
	Module bundler
		Parcel
		Webpack
	Deploying modules with GitHub Pages
	Summary
	Questions
	Further reading
Chapter 11: Tuning Applications to Achieve High Performance
	Technical requirements
	The backend API of TensorFlow.js
		Operations that use the CPU backend
			Low overhead
			Memory management
		Implementing higher parallelism using the WebGL backend
			Avoid blocking the main thread
			Prefer tf.tidy to be free from memory management
			Floating number precision
			Save the overhead of shader compilation
		Using TensorFlow with the Node.js backend
	Tensor management
		Tensor construction
		Tensors as variables
		Revisiting tensor destruction
	Asynchronous data access
	Profiling
		Chrome profiler
	Model visualization
	Summary
	Questions
	Further reading
Chapter 12: Future Work Around TensorFlow.js
	Technical requirements
	Experimental backend implementations
		WebGPU – a new standard for accelerated graphics and computations
		WebAssembly – where the web meets the hardware instruction set
		React Native – moving forward to mobile-native applications
			Creating a React Native app
			Installing the dependencies required for TensorFlow.js
			Writing an application
			Running the application
		Electron – cross-platform desktop environment
	AutoML edge helper
	Summary
	Questions
	Further Reading
Other Books You May Enjoy
Index