WhiteMail
← 用語集

AI・検知エンジン

적대적 예제 (Adversarial Example)

定義

An input subtly perturbed to be misclassified by a model while appearing normal to humans. A core threat in ML security.

詳しく見る

関連用語