YFarmX

Adversarial example

AI

Adversarial example: An input deliberately perturbed, often imperceptibly, to make a model produce a wrong or harmful output while looking normal to humans.

Related terms

Browse the full glossary →