Generating High-Fidelity Synthetic Data for Privacy-Preserving Applications, L. Xu, M. Skoularidou, X. Sun, M. van der Schaar, 2019International Conference on Learning Representations (ICLR)DOI: 10.48550/arXiv.1902.04948 - This paper proposes a GAN-based method for generating synthetic data, focusing on balancing data utility and privacy protection. It illustrates the practical challenges in achieving high fidelity and strong privacy guarantees simultaneously, reflecting the FUP trade-off.
Benchmarking Synthetic Data Generation Models, Y. Lu, S. Ding, R. Huang, F. Wu, 2022ACM Transactions on Knowledge Discovery from Data, Vol. 17DOI: 10.1145/3547271 - This paper provides a systematic benchmark of various synthetic data generation models. It details diverse evaluation metrics for statistical resemblance and machine learning performance, providing a framework for comparing different generation techniques.
The Algorithmic Foundations of Differential Privacy, Cynthia Dwork, Aaron Roth, 2014 (Now Publishers) - This book serves as a principal resource for understanding differential privacy, a method for formally quantifying and preserving privacy in datasets. It lays the theoretical groundwork for privacy protection, which is essential for secure synthetic data generation.