Skip to content

Model Training & Adaptation

Contrastive learning

AI learning useful differences by comparing examples treated as similar or different.

Example

AI learns to connect two differently cropped versions of the same picture.

Why people use it

It teaches AI which examples should be treated as alike or different.

What you'll hear

“These two views should count as the same object.”

What this means for you

Check that examples treated as similar really should be similar for the intended job.

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

Must similar examples look identical?
No. Different crops or lighting can show the same object and still be treated as related examples.
Who decides which examples are similar?
The training setup defines that relationship, sometimes using people and sometimes using automatically created pairs.
Can a bad pairing teach the wrong lesson?
Yes. If different things are treated as alike, useful distinctions can be lost.

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