Defending Against Model Stealing Attacks Using Deceptive Perturbations

Open in new window