LiveFR

Regularization

Definition: Regularization is the set of techniques that constrain a model during training to prevent overfitting and help it generalize better.

It may add a penalty on complexity or randomly drop neurons, for instance. The goal is a simpler, more robust model on unseen data.

See also

← Full AI glossary · AI news