To get this started, here are some of my favourite resources for getting started with Machine Learning in TensorFlow.js. Please do comment below with other resources you have found for ML in JS that helped you out too.
Books:
-
Learning TensorFlow.js by Gant Laborde (Published by O’Reilly): Learning TensorFlow.js [Book]
This book is great for folk who are new to machine learning but familiar with JavaScript and looking to learn the essentials to get started and be productive. From understanding how to manipulate data into Tensors and then quickly progressing to real world applications, this book is a superb introduction that will have you feeling much more confident loading models, passing data to them, and interpreting that data that comes out after digesting these 12 wonderful chapters. Available for pre-order right now (as of March 2021) and having been the technical reviewer for this book I feel this is a very solid introduction for JS devs looking to upskill. -
Deep Learning with JavaScript (Published by Manning): Deep Learning with JavaScript
This book was written by folk on the TensorFlow + TensorFlow.js teams here at Google. It is also aiming to be a solid foundation to the world of Machine Learning and TensorFlow.js and at times gets pretty low level. For me personally this is a solid follow up to read after being familiar with some of the concepts introduced in the O’Reilly book above. From Linear Regression to GANs this book goes deeper in many of these areas.
Courses
Google AI for JavaScript developers with TensorFlow.js by Jason Mayes (Developer Relations Engineer for TensorFlow.js)
Get productive with TensorFlow.js - Google’s Machine Learning library for JavaScript. From pre-made off the shelf models to writing or training your own, learn how to create next gen web apps.
What you’ll learn:
- Common terms and what they mean
- How Machine Learning works (without formal mathematical definitions)
- Overview of the TensorFlow.js library
- Advantages of using ML in JavaScript
- Ways to consume or create Machine Learning models
- How to use pre-made “off the shelf” models
- What Tensors are in Machine Learning
- How to use Tensors with ML models
- How to write a simple custom model
- Perceptrons (artificial neuron) and how they work
- Linear regression to predict numbers using single neuron
- Multi layered perceptrons for handling more complex data
- How to use models that use Convolutional Neural Networks for images
- How to convert Python models to JavaScript
- Transfer learning - reusing existing trained models with your own data
- Inspiring projects others are creating to seed your own future ideas
Browser-based Models with TensorFlow.js by Laurence Moroney
Bringing a machine learning model into the real world involves a lot more than just modeling. This Specialization will teach you how to navigate various deployment scenarios and use data more effectively to train your model.
Building Machine Learning Solutions with TensorFlow.js by Abhishek Kumar
In this course, you’ll learn about the TensorFlow.js ecosystem, and how to set it up on the client-side in the browser and on the server-side with Node.js. First, you’ll discover how to use the environment to build an end-to-end machine learning application that uses natural language processing (NLP) under the hood to detect toxic elements in unstructured text. Next, you’ll learn how to import and process data, build, train, and export a model, and finally predict using the trained model. Finally, you’ll explore how to use existing models trained in Python on the client-side using TensorFlow.js, and even retrain the pre-trained model using transfer learning.
Inspiration
-
TensorFlow.js Community Show & Tell - live every 3 months. Previous episodes after the live event can be found in this YouTube Playlist.
-
#MadeWithTFJS hashtag on social media - a great way to get inspiration and find others who may be working on a similar problem to ask questions to. Right now Twitter and LinkedIn are pretty active with new projects coming out every week from people all around the world:
Twitter Search: https://twitter.com/hashtag/MadeWithTFJS?src=hashtag_click&f=live
LinkedIn Search: Just search the hashtag via the LinkedIn search box when logged in.
Seen other great learning resources? Comment below!