Skip to content

Model Training & Adaptation

Regularization

Ways of guiding AI learning so it relies less on quirks of its practice examples.

Example

A training rule discourages an AI from depending too strongly on particular clues.

Why people use it

It discourages an AI system from fitting every quirk of its teaching examples.

What you'll hear

“Don't learn the accidental details too closely.”

What this means for you

Check that reducing overdependence has not also removed useful patterns.

Can you control it?

Developer-only

The people building or running the AI choose this setup. An everyday user generally needs their help to change how this part works.

Common questions

Does a stronger training penalty always help with new examples?
No. Too strong a penalty can prevent useful patterns from being learned.
Does regularization always mean removing information?
No. It can limit how strongly the system relies on certain patterns or values.
Can it slightly worsen the training score while helping later?
Yes. A less perfect fit to familiar examples can work better on unseen ones.

Related terms

Still have questions?

Up to 500 characters.

Ask LATHIC about AI. Relevant glossary entries may be included.

Your question, the glossary entries it matches, and a rotating pseudonymous identifier go to Microsoft Azure’s OpenAI service through Vercel AI Gateway to generate an answer. Zero retention and no training are required of the provider, and LATHIC does not save your question or answer. Privacy Notice