AI ART Discover · Learn · Create
G AI Glossary Term • Term 88 of 224
Plain English Definition

What does “Generalization” mean?

Core Definition

How well a model performs on new, unseen data outside the training set.

Operational Mechanism

How Generalization Works

Best stopping pointTraining timeErrorError on training dataError on new dataExampleA student who memorizes practice-test answers may fail a new exam, while one who understands the topic can answer new questions. A model that generalizes behaves like the second student on data it has never seen.

Real-World Scenario

Practical Example

A student who memorizes practice-test answers may fail a new exam, while one who understands the topic can answer new questions. A model that generalizes behaves like the second student on data it has never seen.

Core Significance

Why It Matters

The goal of machine learning is good performance on new data, not on the training set. Generalization is what separates a useful model from a memorizer.

Expert Q&A

Common Questions About Generalization

Q: How do I measure generalization?

By testing on held-out data the model has not trained on and comparing with its training performance.

Q: What hurts generalization?

Overfitting, small or unrepresentative training data, and differences between training and real-world conditions.

Q: How can I improve it?

More diverse data, regularization, data augmentation, and simpler models where appropriate.