Evasion Attack

An evasion attack is one of the most common attacks on machine learning models performed during production. It involves designing an input that appears normal to a human but is misclassified by ML models, affecting their behavior; a typical example includes altering pixels in an image to cause an image recognition system to fail in classification.