#native_company# #native_desc#
#native_cta#
  1. Introduction to Machine Learning with TensorFlow.js
  2. Summary
  3. Summary

Summary

Let's get physical

Sometimes, nothing beats holding a copy of a book in your hands. Writing in the margins, highlighting sentences, folding corners. So this book is also available from Amazon as a paperback.

Buy now on Amazon

Congratulations on completing Introduction to Machine Learning with TensorFlow.js. In this final lecture, we’ll quickly review the different lectures, and I will leave you with some final thoughts and where to go next to continue your learning.

Review

Part 1 — Introduction

We talked about the future of JavaScript and Machine Learning and why you might want to learn about the union of these two technologies. I introduced you to the concept of Neural Networks themselves with a simple example and introduced you to TensorFlow and TensorFlow.js. We also covered the setup instructions for all the code samples in this book.

Part 2 — Using a pre-trained model

We built our first application using the pre-trained MobileNet model. We then drilled into the MobileNet model in-depth and discussed how it might take inputs and the actual output of the MobileNet model.

Part 3 — Tensors

Before we can begin to build and train our models, we need to dig into the raw building blocks of TensorFlow, Tensors. We covered how to create them, the operations you can perform on them, and a standard method of calculating the error between two arrays, Mean Squared Error. We then covered TensorFlow.js itself and demonstrated how to use TensorFlow to learn the optimum values for variables in a function.

Part 4 — Regression

Regression is one of the simplest machine learning algorithms you can build and is a great starting point for understanding Neural Networks. We learned how to use TensorFlow.js to construct a linear and polynomial regression model. We used the lower level Core API to get a good understanding of the internals of TensorFlow.js.

Part 5 — Neural Networks

In this chapter, we move towards building a deep neural network. We used the MNIST dataset to construct a model that can predict a hand-drawn digit. We covered a few different Neural Network types from a fully connected dense neural network to a much more complex convoluted neural network that is more suited to work with images.

Part 6 — Transfer Learning

In this final chapter, we pull it together and build a model using a decapitated pre-trained model that we employ in conjunction with another model trained from scratch. Transfer learning requires a lot less training data and a lot less computation, so it is ideal for JavaScript and browser-based work.

Continuing Learning

The purpose of this book is to teach you a practical introduction to Machine Learning. To have at least some skills that you can apply today in your work. If you would like to learn more about Machine Learning, I would recommend a few approached depending on what your overall goal is.

If you want to stick to JavaScript, then a good next book is Deep Learning with JavaScript[1]. It’s written by several people who have worked on the TensorFlow.js product inside Google. It gives you an excellent more in-depth dive into the framework and covers more foundational algorithms.

If you want to keep it practical like this book, then my recommendation is to dissect the other example applications in the TensorFLow.js examples repository[2]. That’s how I learned TensorFlow.js; it’s my favorite method of learning. The example apps are well commented, but there is little in the way of documentation.

I recommend just going ahead and building something, anything, using TensorFlow.js — start using it for something that is the best way to learn how to use it practically.

If this book gave you the bug to turn this into a career for you, then I recommend transitioning over to Python. Most of the other courses and training materials in Machine Learning are in Python, and you’ll find a lot more support. The concepts are the same, so don’t worry, you won’t be starting from scratch.

Two fantastic courses that I recommend taking a look at if you want to pursue this further are:

  • Machine Learning (https://www.coursera.org/learn/machine-learning) This is a famous Machine Learning course by Andrew Ng. He is the co-founder of Coursera and professor at Stanford University. The course is a deep dive into Machine Learning, so be prepared to put much effort into it, but this will give you an excellent bedrock to jump-start your career.

  • Practical Deep Learning for Coders (https://course.fast.ai/) If you would prefer to keep it practical, then this course on fast.ai is a tremendous next step also.

Final Words

I want to thank everyone who supported me, my Kickstarter backers, my wife, and my family. Finishing off this book has been a challenge during COVID-19 but, at times, also a welcome distraction. If you want to stay in contact, please follow me on Linked-in (https://www.linkedin.com/in/jawache/) or Twitter (https://twitter.com/jawache).

Cheers,

Asim


Advanced JavaScript

This unique course teaches you advanced JavaScript knowledge through a series of interview questions. Bring your JavaScript to the 2021's today.

Level up your JavaScript now!