Machine Learning · Chapter 2 of 40
Types of Machine Learning
Three main types: SUPERVISED (learn from labelled examples), UNSUPERVISED (find structure in unlabelled data), and REINFORCEMENT (learn via rewards).
Supervised is by far the most common in industry today.
Example 1 (python)
# Supervised: labeled data (X -> y)
# spam vs not-spam emailsClassic supervised classification.
Example 2 (python)
# Unsupervised: no labels, just X
# customer segmentation via clusteringGroup similar customers automatically.
Key points
- Supervised: predict from labelled examples.
- Unsupervised: find structure in unlabelled data.
- Reinforcement: learn via reward.
- Semi-supervised uses a mix.
💡 Note: AlphaGo (game-playing AI) is a famous reinforcement-learning success.