10. What is Machine Learning?

Separating Data

When we talk about machine learning, you’ll often hear the terms “deep learning” and “neural networks” come up. This might conjure up an image of a brain or some strange graphic of layers of mathematical formulas and data. But at their core, all of these learning techniques are about separating data into different classes.

The image that this conjures in my head is of a child playing on a beach. The child sees some blue and yellow shells in the sand.

Then, someone says to the child: “put these shells into groups and draw a line between them.”

Without me telling you anything else, how would you group these shells?

You might separate them by color and shape and draw a line between these two types of shells. Now, for a computer, the scenario might be that we give it individual images of shells, and just like a child, a neural network can learn how to separate these images of shells based on similarities or differences in the given examples.

After this separation step, if a network sees a new image, one it hasn’t seen before, it sorts and classifies it based on which side of the line it falls!

Realistically, data is often a lot more complex than this, but neural networks just layer separation on top of separation layer to create more complex boundaries and group all kinds of data!