Unsupervised Learning: Explained, Briefly

“You can have data without information, but you cannot have information without data.” — Daniel Keys Moran

Unsupervised learning refers to machine learning algorithms that learn through unlabelled data. For instance, images of various images of animals would be inputted and the algorithm will start looking for similar patterns and start grouping similar images together.

Let’s break it down further:

1. We start by inputting raw data (the unlabelled animals)
2. We first tell the number of groups we want the algorithm to form. (2 groups in this case)
3. In this step, the algorithm starts to look for similarities between each image and then begins to group.

How perfect… not

Like everything, even unsupervised learning isn’t flawless. There are a few problems that may restrict us from using this type of algorithm.

→ Because unsupervised learning begins from raw data, it’s highly time-consuming because there are endless opportunities.

→ Since there are no labels, it’s also prone to having higher chances of inaccuracy.

Unsupervised Learning or Supervised Learning?

When it comes to comparing unsupervised and supervised learning, the choice depends on the situation. In supervised learning, it’s required for you to have input and output data. Whereas unsupervised learning does not require output data; however, it is highly time-consuming and not so accurate. The choice is yours!