Skip to content

The code for the book Learning TensorFlow.js by Gant Laborde - Published by O'Reilly Media

Notifications You must be signed in to change notification settings

chinh-le/learn-tfjs

 
 

Repository files navigation

Source for O'Reilly's Learning TensorFlow.js Book

Learning TensorFlow.js - Powerful Machine Learning in JavaScript by Gant Laborde

book cover

About the Book

Learn how to take advantage of the TensorFlow.js framework to implement machine learning models in the client browser or server.

This book is intended for two audiences:

  • Web devs and Front end Engineers who are familiar with JavaScript but unfamiliar with how to get started in AI / ML.
  • Experienced AI specialists who are interested in how to apply their server-based skills to a framework like TensorFlow.js.

Purchase your copy of the book on Amazon: https://amzn.to/3dR3vpY

About the Code

The code in this repository is broken down by chapter. Each chapter folder has a technical domain split. Some code is repeated in each folder for each technology.

The folders in each chapter could be:

  • extra - any extra content for that chapter that is not technically specific.
  • node - A Node.js set of solutions and code for the given chapter that run as a server.
  • simple - A "inline" hosted set of HTML solutions in code for a given chapter that run in the browser. These files do not depend on a package management system for hosting. These files access their dependencies via CDNs.
  • web - A Parcel.js web hosted solution of code that runs using NPM to create a browser based solution. These projects reflect modern transpiled web technology.

Book Chapters

  • Chapter 1 AI is Magic - There is no code associated with Chapter 1 because it's an introduction to the book and concepts. I've added a small readme with some of the links mentioned in the chapter for convenience.
  • Chapter 2 Introducing TensorFlow.js - This chapter is focused on getting you running TensorFlow.js on a client or a server. Once you've got it running, you actually run a Toxicity classifier on given text.
  • Chapter 3 Introducing Tensors - This chapter helps you understand the concept and need of tensors. You then immediately use this technology to build a simple recommendation system for music.
  • Chapter 4 Image Tensors - Images in machine learning are a fantastic example of tensors and all the things you can do to modify complex data.
  • Chapter 5 Introducing Models - Learn what makes an AI tick. Machine learning models are the core of what drives machine learning. In this chapter, you implement several models.
  • Chapter 6 Advanced Models & UI - In this chapter, you implement a very advanced model that detects objects, you then do an overlay that helps illustrate the results, and you connect everything to the webcam for real-time inference.
  • Chapter 7 Model Making Resources - Now that you understand how to implement models, where do they come from? This chapter gives you a tour of conversion commands and data resources.
  • Chapter 8 Training Models - Train your first model from data. See the simplest model architecture for the simplest problem. You train directly in the browser!
  • Chapter 9 Classification Models & Data Analysis - Data isn't always clean. Learn how to build a notebook, visualize, and extract features from your data by solving who would survive the Titanic.
  • Chapter 10 Image Training - Bring in some advanced concepts for feature extraction via convolutions. Understand and learn how to build more advanced models on Node.js and implement those models in the browser.
  • Chapter 11 Transfer Learning - Learn what transfer learning is and utilize it. Transfer learn with several methods and see the benefit with small datasets.
  • Chapter 12 Dicify - Capstone Project - Utilize all the skills you've learned. Compose a dataset and train a model to create art out of dice.

About

The code for the book Learning TensorFlow.js by Gant Laborde - Published by O'Reilly Media

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • HTML 44.3%
  • JavaScript 42.0%
  • TypeScript 12.9%
  • CSS 0.8%