Learning TensorFlow.js

Chapter 2. Introducing TensorFlow.js

“If your actions inspire others to dream more, learn more, do more, and become more, you are a leader.”

—John Quincy Adams

We’ve been talking about TensorFlow.js a bit and what it can do, but we haven’t really dug into what a machine learning framework like TensorFlow.js actually is. In this chapter, we’ll tackle the concept of a machine learning framework and then quickly dive into writing code. I know it’s important to write code that has some kind of tangible outcome, so in this chapter, you’ll finally get your computer running TensorFlow.js and producing results.

We will:

• Look at the concept of TensorFlow.js

• Set up TensorFlow.js

• Run a TensorFlow.js model package

• Take a deep look at what the AI did

Let’s start with the framework we’ll be using to make it all happen.

Hello, TensorFlow.js

Given how our previous chapter discussed the philosophies of ancient times and the birth of machine learning as a field, you’d expect AI frameworks to have a history that reaches as far back as the early 1960s. However, AI was stagnant for a long time, and this time is often called “AI winter.” The concepts of AI were beleaguered by disbelief and extreme mathematical calculations for the small data that was available. Who could blame these researchers? Most software developers today depend on shipping apps without writing GPU-enabled linear algebra and calculus from scratch, and building your own AI shouldn’t be the exception. Fortunately, due ...