The paper introduces Jacobian-Guided Anisotropic Noise Reshaping, a method that improves data utility under Local Differential Privacy by selectively reducing noise in task-relevant subspaces of data representations. It uses the Jacobian of a public downstream model to identify critical directions and reshapes isotropic LDP noise into an anisotropic distribution, preserving privacy while enhancing performance. Experiments on CIFAR-10-C show significant accuracy gains, especially for PrivUnit variants at ε=7.5.
arXiv:2508.04800v2 Announce Type: replace-cross Abstract: We introduce a novel privatization framework for high-dimensional controlled variable selection. Our framework enables rigorous False Discove...
arXiv:2602. 06838v3 Announce Type: replace Abstract: Federated learning enables collaborative model training across distributed clients while preserving data privacy.
The paper investigates high‑dimensional LASSO under differential privacy using objective perturbation when covariates have heterogeneous scales. It introduces a Gram‑based anisotropic objective perturbation that counteracts the distortion caused by covariate heterogeneity, restoring isotropy in the estimation process. Through an Approximate Message Passing framework and state evolution analysis, the authors show that this approach stabilizes convergence and improves both statistical efficiency and privacy performance compared to standard uniform noise injection.
The paper introduces Differential Privacy Representation Geometry for Medical Imaging (DP‑RGMI), a framework that interprets differential privacy as a structured transformation of representation space. DP‑RGMI decomposes performance loss into encoder geometry—measured by representation displacement and spectral effective dimension—and task‑head utilization, quantified by the gap between linear‑probe and end‑to‑end utility. Across 594,000 chest X‑ray images from four datasets, the study finds that differential privacy consistently creates a utilization gap even when linear separability remains, while displacement and spectral dimension vary non‑monotonically with initialization and dataset, indicating that privacy alters representation anisotropy rather than uniformly collapsing features.
arXiv:2606. 17995v1 Announce Type: cross Abstract: We study the privacy of releasing posterior sample paths from a Gaussian process (GP) when the entire training set including covariates and responses is private.
The paper investigates how differential privacy (DP) affects the ability to merge independently fine‑tuned task models into a single multi‑task model. It identifies two geometric challenges—local sharpness and reference drift—that hinder mergeability of private models. To address these, the authors propose DP‑Merging, a framework that uses a sharpness‑aware objective and a reference‑based alignment regularizer to improve mergeability while maintaining DP guarantees.
arXiv:2607. 05866v1 Announce Type: cross Abstract: Under a fixed privacy budget, the utility of differentially private (DP) training is ultimately determined by its optimization efficiency.
arXiv:2405. 16361v4 Announce Type: replace Abstract: To protect privacy in regulated domains such as healthcare and finance, model owners may allow only remote API access while keeping both the training data and model parameters private.
arXiv:2608. 05737v1 Announce Type: cross Abstract: Local Differential Privacy (LDP) provides strong privacy guarantees for collecting numerical data.
arXiv:2601. 17360v2 Announce Type: replace-cross Abstract: An adversary observing a model's released prediction can infer sensitive attributes of the queried input, or even reconstruct representatives of the model's training data.
arXiv:2606. 04399v1 Announce Type: new Abstract: In the paradigm of decentralized learning, a group of agents collaborate to train a global model using distributed datasets without a central server.
arXiv:2601. 11219v3 Announce Type: replace-cross Abstract: Federated learning (FL) for large language models (LLMs) has attracted increasing attention as a privacy-preserving approach for adapting models over distributed data, where parameter-efficient methods such as Low-Rank Adaptation (LoRA) are widely adopted to reduce communication and memory costs.