Jacobian-Guided Anisotropic Noise Reshaping for Enhancing Representation Utility under Local Differential Privacy

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Ha, Youngmok, Schlegel, Viktor, Sun, Yidan, Bharath, Anil Anthony
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917513930997760
author Ha, Youngmok
Schlegel, Viktor
Sun, Yidan
Bharath, Anil Anthony
author_facet Ha, Youngmok
Schlegel, Viktor
Sun, Yidan
Bharath, Anil Anthony
contents While Local Differential Privacy (LDP) serves as a foundational primitive for distributed data collection, its stringent noise injection requirement often leads to severe degradation in data utility. This degradation stems from the task-agnostic nature of conventional LDP mechanisms, which inject noise uniformly across all dimensions regardless of their relative importance to the downstream objective. To address this issue, we propose a novel approach that mitigates noise in task-relevant subspaces of the data representation. Our method identifies task-critical subspaces via the Jacobian matrix of the public downstream model, selectively attenuates noise along those dimensions, and reshapes the isotropic noise of standard LDP into an anisotropic distribution. This method preserves the uniform per-dimension privacy budget while heterogeneously modulating noise impact across dimensions, thereby substantially enhancing data utility. Furthermore, our approach generalizes to both linear and non-linear models and integrates seamlessly with existing mechanisms. Extensive experiments on CIFAR-10-C (Brightness corruption at the highest severity level 5) demonstrate that integrating our approach improves the utility of PrivUnit2 and PrivUnitG by approximately 20\% at $ε=7.5$. The source code is available at https://github.com/ymha/jacobian-anr-ldp.
format Preprint
id arxiv_https___arxiv_org_abs_2605_16812
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Jacobian-Guided Anisotropic Noise Reshaping for Enhancing Representation Utility under Local Differential Privacy
Ha, Youngmok
Schlegel, Viktor
Sun, Yidan
Bharath, Anil Anthony
Machine Learning
Cryptography and Security
While Local Differential Privacy (LDP) serves as a foundational primitive for distributed data collection, its stringent noise injection requirement often leads to severe degradation in data utility. This degradation stems from the task-agnostic nature of conventional LDP mechanisms, which inject noise uniformly across all dimensions regardless of their relative importance to the downstream objective. To address this issue, we propose a novel approach that mitigates noise in task-relevant subspaces of the data representation. Our method identifies task-critical subspaces via the Jacobian matrix of the public downstream model, selectively attenuates noise along those dimensions, and reshapes the isotropic noise of standard LDP into an anisotropic distribution. This method preserves the uniform per-dimension privacy budget while heterogeneously modulating noise impact across dimensions, thereby substantially enhancing data utility. Furthermore, our approach generalizes to both linear and non-linear models and integrates seamlessly with existing mechanisms. Extensive experiments on CIFAR-10-C (Brightness corruption at the highest severity level 5) demonstrate that integrating our approach improves the utility of PrivUnit2 and PrivUnitG by approximately 20\% at $ε=7.5$. The source code is available at https://github.com/ymha/jacobian-anr-ldp.
title Jacobian-Guided Anisotropic Noise Reshaping for Enhancing Representation Utility under Local Differential Privacy
topic Machine Learning
Cryptography and Security
url https://arxiv.org/abs/2605.16812