Advances in Differential Privacy Mechanisms

The field of differential privacy is rapidly evolving, with a growing focus on developing innovative mechanisms to protect sensitive data. Recent research has explored new approaches to achieving differential privacy, including the use of symmetric alpha-stable distributions and conformal transformations on Riemannian manifolds. These advancements have the potential to provide more accurate and private data analysis, particularly in scenarios where data is distributed across multiple sources or has complex geometric structures. Noteworthy papers in this area include the Symmetric alpha-Stable mechanism, which achieves pure differential privacy while remaining closed under convolution, and Conformal-DP, which utilizes conformal transformations to equalize local sample density and redefine geodesic distances. Additionally, the development of verifiable distributed differential privacy mechanisms, such as VDDP, has addressed the need for trustworthy execution of differential privacy mechanisms in distributed settings.

Sources

Heavy-Tailed Privacy: The Symmetric alpha-Stable Privacy Mechanism

Bipartite Randomized Response Mechanism for Local Differential Privacy

Conformal-DP: Differential Privacy on Riemannian Manifolds via Conformal Transformation

An Inversion Theorem for Buffered Linear Toeplitz (BLT) Matrices and Applications to Streaming Differential Privacy

VDDP: Verifiable Distributed Differential Privacy under the Client-Server-Verifier Setup

On the Space Complexity of Online Convolution

Built with on top of