AIIntermediate

Overfitting

Learning training-specific details so strongly that performance suffers on new data.

In the real world

A model fits training noise but predicts poorly on new examples.

Related

Go a little deeper

Google ML Glossary (opens in a new tab)

Improve this entry