Can we infer the presence of Differential Privacy in Deep Learning models' weights? Towards more secure Deep Learning

Open in new window