Machine learning is a component of AI – the part that allows AI technology to learn from past experiences, or data it previously processed.

Machine learning can happen in either supervised or unsupervised ways.

Supervised machine learning can make predictions based on existing data. A website that recommends certains products might, for instance, be using supervised machine learning.

But how exactly does it happen?

How does it get from lists of data to predictions about what you’ll want to buy next?

It uses algorithms – quite a few of them. One of these is the classification algorithm.

This algorithm is useful for automatically categorising items in a database. It is, for example, used to predict whether you’ll mark an email you received as spam or not.

The media update team breaks this algorithm down to the basics:

So what exactly is classification?

Classification in machine learning is when a machine or computer uses an algorithm to draw conclusions from data that it already has, and then uses these conclusions to categorise new data it receives.

Writing for AI Zone, Sarah Davis puts it this way: “Classification algorithms let machines assign a category to a data point based on training data.”

The inner workings of spam filters is an easy way to explain how classification works:
  • A software developer creates a database that contains emails as well as a field to describe whether each of these emails are spam or not. 
  • The classifier algorithm then learns from the information in the database when an email is spam. It creates a model based on what it has learnt. 
  • It then applies its model to new emails and predicts whether an email is spam or not. 
In this example, the classification algorithm can only determine one of two answers. That’s what makes classification different from other supervised learning algorithms – like the regression algorithm.

While regression algorithms are also used to predict an outcome, they do so for specific values such as a stock price. Its predictions aren’t only one of a few values that the IT developer has defined in a database.

That’s the basics of classification.

The media update team has also written about other AI terms. Learn more with these:
Or read our article on Three reasons why AI is a marketer’s best friend.
Image designed by Freepik