Defending adversarial attacks

In adversarial machine learning, defense refers to Machine Learning techniques that attempt to protect models from being attacked by malicious attempts.