GlossarIA
Open AI glossary for companies
← Back to glossary

Adversarial attacks

Techniques that manipulate inputs in ways imperceptible to humans but that fool the model.

Advanced seguridad ataques robustez

Full definition

Techniques that manipulate inputs in ways imperceptible to humans but that fool the model.

Example in a business context

Adding noise to a stop sign image so the model classifies it as speed limit.