🎓 The Three Types of Learning

Machine learning isn't one method. There are three main styles, and each fits a different kind of problem. Knowing which is which is one of the most useful things a beginner can learn — it shows up in interviews, courses, and every real project.

1. Supervised Learning — Learning With an Answer Key

You give the model labeled examples: input + correct answer. It learns to map inputs to outputs. This is the most common type in practice.

• •Classification — predict a category: spam / not spam, cat / dog, fraud / legitimate
• •Regression — predict a number: house price, tomorrow's temperature, expected sales
• •Needs lots of labeled data, which can be expensive to create (someone has to label it)

2. Unsupervised Learning — Finding Hidden Structure

You give the model data with NO labels and ask it to find patterns or groupings on its own. It doesn't know the 'right answer' — there isn't one.

• •Clustering — group similar things together: customer segments, similar documents
• •Anomaly detection — flag things that don't fit the pattern (hugely important in cybersecurity!)
• •Great when you don't have labels — which is most real-world data

3. Reinforcement Learning — Learning by Trial and Reward

An 'agent' takes actions in an environment and gets rewards or penalties. Over time it learns a strategy that maximizes reward. This is how AI mastered games like Go and how robots learn to walk.

• •No dataset — the agent generates its own experience by trying things
• •Powered breakthroughs like AlphaGo beating the world Go champion in 2016
• •Used in robotics, game AI, and increasingly to fine-tune chatbots (RLHF)