TensorFlow.js Deep Learning Projects

More Information
  • Understand the TensorFlow.js API and its usage in the deep learning domain
  • Explore model training and data visualization in TensorFlow.js
  • Create a sentimental analysis web app with multiple models
  • Generate interesting synthetic images using GANs
  • Implement a self-balancing simulation web app using reinforcement learning
  • Build exciting self-learning games using multiple deep learning neural network architectures
  • Detect poses in real-time using PoseNet
  • Develop interesting effects such as making parts of an image invisible in a frame with the BodyPix model

TensorFlow.js is an open source machine learning (ML) library from Google for performing browser-based ML. It enables developers to create, run, and embed complex ML models in the browser. With this book, you’ll learn to integrate autonomous and faster analytical capabilities in the browser efficiently.

The book starts from the basics, explaining the fundamentals of TensorFlow.js and machine learning, before building up to the advanced concepts. You’ll understand how to develop and deploy lightweight ML models. The book then takes you through developing smart apps directly on web browsers by using deep learning and TensorFlow.js features. You’ll not only learn to implement neural network architectures such as RNNs, GRUs, and LSTMs, but also work through end-to-end projects, right from creating a speech recognizer and web app for text generation, to making a synthetic image generator system and simulation web app. Covering all the steps involved, from model building to deployment, along with best practices, you’ll gain plenty of practical experience. Finally, you’ll use multiple neural network architectures to build a game that can re-train your model with multiple pictures of gestures.

By the end of this deep learning book, you’ll have knowledge of TensorFlow.js and browser-based ML and be able to easily embed ML in your JavaScript projects.

  • Build lightweight machine learning models and run them directly on the browser
  • Become proficient in machine learning tasks to make your web applications dynamic and analytics-driven
  • Create advanced neural networks, train pre-trained models and get to grips with TensorFlow.js features
Page Count 370
Course Length 11 hours 6 minutes
ISBN 9781789538328
Date Of Publication 5 Feb 2021


Mr Umang Sharma

Umang Sharma has been a National Science Academies Summer Research Fellow in 2014 worked on an Astrophysics machine learning project, He has also worked on a CERN project, later joint a tech startup leading the A.I practice. Currently, he works for Fortune 100 companies building data science solutions. Apart from day to day work, he is a keen open-source contributor. He has contributed to Google Datalab, built a TensorFlow.js IntelliJ IDE plugin and few more. He is also a prominent speaker in machine learning and deep learning. He has been the official speaker of Google’s machine learning crash course and Google DevFest 2018 and a large number of National Tech Conferences and AI fests. He also got featured on Google’s machine learning crash course official newsletter multiple times.