Unsupervised learning
Finding structure in data with no labels — the model discovers patterns on its own.
Think of it like
Like sorting a pile of mixed photos into natural groups without anyone telling you the categories.
Example
A clustering algorithm groups customers into segments from their behavior, with no predefined labels to learn from.
How it actually works
Unsupervised learning uncovers hidden structure — clusters, densities, low-dimensional representations — without target outputs. Classic examples are k-means clustering and dimensionality reduction. It overlaps with self-supervised learning, but the distinction is subtle: self-supervision manufactures prediction targets from the data, while pure unsupervised methods seek structure without any prediction target at all.
For product teams
Useful when you have lots of data but no labels — it finds groupings and structure to act on.
For engineers
Learning structure (clusters/manifolds) without labels or targets; distinct from self-supervised prediction tasks.
Related
- Supervised learning — Contrast with supervised learning.
- Self-supervised learning — Overlaps with self-supervised learning.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome