Stateful Defenses for Machine Learning Models Are Not Yet Secure Against Black-box Attacks

Open in new window