Article: Tensors Illustrated (Hacker Noon)
Daniel Jeffries’ relatively math-free explainer on Tensors…with cats
As someone new to AI, you’ll start coming across terms like TensorFlow, Theano, Torch, PyTorch etc. While you may appreciate that these are machine learning libraries and frameworks, it’s probably useful to understand the absolute basics. In this case, what is a tensor?
Daniel walks you through the basics:
Tensors = Containers
A tensor is the basic building block of modern machine learning.
At its core it’s a data container. Mostly it contains numbers. Sometimes it even contains strings, but that’s rare.
So think of it as a bucket of numbers.
Most of his walk-through describes successive dimensions of these tensor building blocks, from 0D (aka scalars) through to 5D (like sets of videos). He uses examples throughout, like the one below for an image of his cat.
By the end, I better understood how different data inputs could be represented as tensors, which is a step in the right direction for me.
Daniel’s piece here is actually one of a five-part series on Learning AI If You Suck at Math. Here are the other four parts, which I’d recommend reading if you’re fond of his tensor explainer. Enjoy!
