Mar 7, 2024 · Experiments on model training with DP-SGD show that using bounded support Gaussian mechanisms can provide a reduction of the pDP bound \epsilon ...
Mar 7, 2024 · In this work we proposed a novel analysis for two privacy mechanisms that consider bounding the support set for the Gaussian mechanism. We ...
Repo for the project "Privacy Amplification for the Gaussian Mechanism via Bounded Support" - facebookresearch/bounded_gaussian_mechanism.
The authors propose modifications to the Gaussian mechanism with bounded support to amplify privacy guarantees under data-dependent accounting, ...
Feb 2, 2024 · Finally, we introduce a new family of mechanisms based on diffusion processes which are closed under post-processing, and analyze their privacy via a novel heat ...
Feb 11, 2024 · The Gaussian mechanism is one differential privacy mechanism commonly used to protect numerical data. However, it may be ill-suited to some ...
Missing: Amplification via
Abstract. In the past decade, differential privacy has seen remark- able success as a rigorous and practical formalization of data privacy.
We further prove that the RDP is strictly upper-bounded by the Gaussian RDP with- out shuffling. The shuffle Gaussian RDP is advantageous in composing ...
In this paper, we provide the first general result of “privacy- amplification” of RDP via Poisson subsampling. Our main contributions are the following. 1.
A fundamental result in differential privacy states that the privacy guarantees of a mechanism are preserved by any post-processing of its output.