Unsupervised learning

Involves training a model on an unlabeled dataset. The goal is to discover hidden patterns or structures in the data. It is like giving a child a big box of LEGO bricks without the instruction manual. The goal is to see what interesting structures or patterns they come up with on their own. Examples include clustering and dimensionality reduction tasks:

• Clustering: The goal is to group similar instances into clusters. It's like organizing your toys into groups without being told how. For example, a shop might group customers into different types so they can figure out the best way to advertise to each type. Or, on social media, it helps in finding who's in which friend group.
• Dimensionality Reduction: A technique that simplifies the complexity of high-dimensional data while retaining its essential characteristics. Imagine if you have a huge backpack full of school supplies but you want to make it lighter. Dimensionality reduction is like figuring out what's essential and what's not, so you only carry what you really need. This helps in dealing with too much information - like if you're trying to look at a very complex picture and simplify it so it's easier to understand.

As we go through different algorithms in this course, we'll talk about whether they're more like supervised or unsupervised learning.