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

Abstract

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

Differential Advising in Multiagent Reinforcement Learning

“…Differential privacy is a prevalent privacy model and has been broadly applied to various applications [20], [21], [22]. Differential privacy can guarantee an individual's privacy independent of whether the individual is in or out of a dataset [7].…”

An Improved Genetic Algorithm for Safety and Availability Checking in Cyber-Physical Systems

“…The separation of duty (SOD) policies is a typical policy used to ensure safety [19]. It prevent a set of users less than a certain threshold from being fully authorized to perform sensitive tasks [15], [16], [20]. Excessive pursuit of system safety may lead to unavailability of the system.…”

Privacy preservation for image data: A GAN‐based method

“…Zhu et al have pioneered the use of differential privacy in several quarters of AI, such as multiagent systems, reinforcement learning, and knowledge transfer. 19,20 In terms of image privacy preservation, Fan was the first to extend the concept of differential privacy with a scheme of sharing pixelized images, 21 while Yuan et al applied differential privacy to medical images. 22 In our framework, we use GANs are an efficient tool to remove sensitive information without destroying the original image or losing its salient attributes, which sets our solution apart from obfuscation methods.…”