Secure and Private Machine Learning: A Survey of Techniques and Applications
Raja Abou · 2023
Machine Learning (ML) privacy violations can lead to discrimination and identity theft, among other serious repercussions. As more sensitive data has been utilized to train models in recent years, the necessity for privacy-preserving methods in ML has grown in significance. The state-of-the-art methods for Privacy-Preserving Machine Learning (PPML), such as safe multi-party computation, homomorphic encryption, and differential privacy, are thoroughly reviewed in this survey study. We also assess PPML’s drawbacks, including scalability, computing efficiency, and the trade-off between privacy and utility. Finally, we identify the open problems and future directions of research in PPML, including emerging trends, challenges, and opportunities. This survey paper is intended to serve as a valuable resource for researchers and practitioners interested in the area of PPML.