Chapter 04 · Section I · 12 min read
The types of machine learning
Supervised, unsupervised, reinforcement — three modes of learning, explained with examples you can hold in your hand.
There are three broad ways to teach a machine.
Supervised learning. You show it many examples, each with the correct answer attached. Khalti’s fraud model is supervised: every historical transaction is labelled “fraud” or “not fraud”, and the model learns the difference.
Unsupervised learning. You give it data with no labels and ask it to find structure. Group these 50,000 Pathao trips into “kinds of trip”; you didn’t tell the model what the kinds are, but it can find them.
Reinforcement learning. The model takes actions, gets rewards, and updates its strategy. This is how AlphaGo learned to play Go — by playing itself millions of times.
This section is a stub. The full version will work through one example of each, using Nepali data.