PROSAC: Provably Safe Certification for Machine Learning Models under Adversarial Attacks

Open in new window