A system stores data without training a model

A model discovers hidden groups without target labels.

An agent learns only through rewards and penalties.

A model learns from labeled input-output examples.