Backend
Federated Differential Privacy with Batch Contamination
arXiv stat.TH0 views
arXiv:2611.23551v1 Announce Type: new Abstract: We study mean estimation under federated differential privacy (FDP) constraints and batch contamination, considering both Gaussian and heavy-tailed distributions. In the homogeneous setting, where sample sizes and privacy budgets are constant across servers, we show that the costs of privacy and contamination disentangle. Under a bounded $v$th moment condition, with $v \in (1, \infty)$, the cost of batch contamination is of order $(\varepsilon/n)^{2-2/v}\vee \varepsilon^2/n$, where $\varepsilon$ is the contamination level and $n$ is the per-batch sample size. This reveals a phase transition at $\varepsilon\asymp n^{1-v/2}$, for $v > 2$, which is the first of its kind in the literature. We develop a two-stage procedure that achieves the minimax rate up to poly-logarithmic factors. A key component is a private smoothed voting procedure tailored to the FDP setting that provides optimal sensitivity control. For heterogeneous batch sample sizes and privacy budgets, we introduce a batch partitioning framework that groups batches with comparable per-batch sample sizes and privacy budgets. Surprisingly, the minimax rate is determined, up to poly-logarithmic factors, by the most informative group of batches. Consequently, in some regimes, it is minimax optimal to use only an arbitrarily small proportion of the data set. Heterogeneity also introduces interactions between the costs of privacy and contamination, in contrast to their disentanglement in the homogeneous setting. For unknown contamination level, we propose an adaptive, near-minimax optimal Lepski-type extension for the heterogeneous Gaussian setting, designed specifically to prevent privacy leakage by ensuring no repeated use of data points.
Source:https://arxiv.org/abs/2611.23551PreviousMinimax estimation for simultaneous deconvolution model in random design using waveletsNextSpectral Stability and Convergence of the Unnormalized Graph Laplacian on Noisy Point Clouds
Related posts
PyPy: First Contact with RPython
0Hacker News New
Show HN: Chipload – Open-source 2.5D CAM for CNC routing
0Hacker News Show
Show HN: Openspot – macOS launcher for disposable tools
0Hacker News Show
Insights into data center resource use in Nevada
0Phys.org Tech
AI helps uncover critical flaw in ubiquitous internet software
0Phys.org Tech
Fast and furious: NATO wrestles with speed of AI-driven warfare
0Phys.org Tech