Privacy Preserving
Privacy-preserving techniques aim to enable data analysis and machine learning while safeguarding sensitive information. Current research focuses on developing and improving methods like differential privacy, federated learning, homomorphic encryption, and data obfuscation, often applied to specific model architectures such as transformers and neural radiance fields. These advancements are crucial for addressing privacy concerns in various applications, including healthcare, finance, and AI-powered services, allowing for collaborative data analysis and model training without compromising individual privacy. The field is actively exploring the trade-offs between privacy guarantees, model accuracy, and computational efficiency.
Papers
Differentially Private Graph Neural Network with Importance-Grained Noise Adaption
Yuxin Qi, Xi Lin, Jun Wu
FaceSkin: A Privacy Preserving Facial skin patch Dataset for multi Attributes classification
Qiushi Guo, Shisha Liao
Collaborative Learning From Distributed Data With Differentially Private Synthetic Twin Data
Lukas Prediger, Joonas Jälkö, Antti Honkela, Samuel Kaski
When Federated Learning meets Watermarking: A Comprehensive Overview of Techniques for Intellectual Property Protection
Mohammed Lansari, Reda Bellafqira, Katarzyna Kapusta, Vincent Thouvenot, Olivier Bettan, Gouenou Coatrieux
Binary Federated Learning with Client-Level Differential Privacy
Lumin Liu, Jun Zhang, Shenghui Song, Khaled B. Letaief