Verification of Neural Network Control Policy Under Persistent Adversarial Perturbation
2019·Arxiv