Defending Against Deepfakes with Ensemble Adversarial Perturbation | IEEE Conference Publication | IEEE Xplore