Core ML Concepts · intermediate · concept 37 of 311 · Updated Nov 4, 2026
Active Learning
Active learning lets a model choose which examples a person should label next, usually the ones it is least sure about. When labels are expensive, as with medical images or legal documents, this can reach the same accuracy with far fewer of them. It needs a labeling workflow people can actually sit inside, which is often the harder part.
Key terms
Uncertainty samplingQuery strategyLabeling budgetHuman in the loopPool-based learning
Learn these first
Where you meet it in the real world
Medical image labeling, document review, improving classifiers from real errors
Guides and articles