Model Poisoning

Adversarial Machine Learning: How Attackers Can Fool AI Security Systems

As security products rely more on machine learning, attackers are learning to target the models themselves. Tiny, carefully chosen changes c…