poisoning attacks

Adversarial attacks in which an adversary interferes with a model during its training stage, such as by inserting malicious training data (data poisoning) or modifying the training process itself (model poisoning).

Український переклад ще не підготовлено.

NIST AI 100-2e2025