AI for networks: graph machine learning
Graph learning uses patterns between connected things, not just within individual rows of data.
What is it?
Graph machine learning learns from networks of connected units: people, molecules, roads or transactions. It looks not only at a single example but at what it is connected to. Often the relationship is what reveals the pattern.
An everyday example
A bank spots a suspicious ring of accounts passing money around in a circle, even though each transfer looks ordinary on its own.
Why does it matter?
It helps with fraud detection, drug discovery, recommendations and traffic planning, where links matter more than individual records.
A common misconception
That a link means cause or guilt. Being in someone’s network is not proof that you did the same thing.
In more detail
What is an example?
Predicting the properties of molecules, analysing networks or uncovering linked fraud.
What is a graph neural network?
A model that combines information from points in a network (nodes) and their neighbours.
Why can it work better?
Because the relationships between things sometimes matter more than any single property.
Where can it go wrong?
When the links are incomplete or misleading, or when the environment changes.
Remember: Not every piece of data stands alone: sometimes the relationship is the point.
Reliable sources
Last reviewed 1 November 2026. An explainer, not professional, legal or medical advice. For rules and figures that change, check the official version.