More Than Privacy: Applying Differential Privacy in Key Areas of Artificial Intelligence

Artificial Intelligence (AI) has attracted a great deal of attention in\nrecent years. However, alongside all its advancements, problems have also\nemerged, such as privacy violations, security issues and model fairness.\nDifferential privacy, as a promising mathematical model, has several attractive\nproperties that can help solve these problems, making it quite a valuable tool.\nFor this reason, differential privacy has been broadly applied in AI but to\ndate, no study has documented which differential privacy mechanisms can or have\nbeen leveraged to overcome its issues or the properties that make this\npossible. In this paper, we show that differential privacy can do more than\njust privacy preservation. It can also be used to improve security, stabilize\nlearning, build fair models, and impose composition in selected areas of AI.\nWith a focus on regular machine learning, distributed machine learning, deep\nlearning, and multi-agent systems, the purpose of this article is to deliver a\nnew view on many possibilities for improving AI performance with differential\nprivacy techniques.\n