LDP-Slicing: Local Differential Privacy for Images via Randomized Bit-Plane Slicing

2026-03-03

AI Summary

This work addresses the severe utility degradation of local differential privacy (LDP) in high-dimensional pixel spaces of images. To this end, the authors propose LDP-Slicing, a lightweight and training-free framework that, for the first time, effectively achieves pixel-level ε-LDP protection for images. The approach decomposes pixels into binary bit-planes, applies LDP at the bit level, and integrates a perceptual obfuscation module with an optimized privacy budget allocation strategy. This design substantially mitigates the utility loss inherent in high-dimensional settings. Empirical evaluations demonstrate that LDP-Slicing outperforms existing DP and LDP methods on face recognition and image classification tasks, maintaining high downstream performance under the same privacy budget while incurring minimal computational overhead.

Technology Category

• Machine Learning: Privacy
• Computer Vision: Bias, Fairness & Privacy
• Search and Optimization: Learning to Search

Application Category

• Security and Privacy: Privacy-enhancing technologies
• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Local Differential Privacy (LDP) is the gold standard trust model for privacy-preserving machine learning by guaranteeing privacy at the data source. However, its application to image data has long been considered impractical due to the high dimensionality of pixel space. Canonical LDP mechanisms are designed for low-dimensional data, resulting in severe utility degradation when applied to high-dimensional pixel spaces. This paper demonstrates that this utility loss is not inherent to LDP, but from its application to an inappropriate data representation. We introduce LDP-Slicing, a lightweight, training-free framework that resolves this domain mismatch. Our key insight is to decompose pixel values into a sequence of binary bit-planes. This transformation allows us to apply the LDP mechanism directly to the bit-level representation. To further strengthen privacy and preserve utility, we integrate a perceptual obfuscation module that mitigates human-perceivable leakage and an optimization-based privacy budget allocation strategy. This pipeline satisfies rigorous pixel-level $\varepsilon$-LDP while producing images that retain high utility for downstream tasks. Extensive experiments on face recognition and image classification demonstrate that LDP-Slicing outperforms existing DP/LDP baselines under comparable privacy budgets, with negligible computational overhead.

Problem

• Local Differential Privacy
• Image Privacy
• High-Dimensional Data
• Utility Degradation
• Pixel-Level Privacy

Innovation

• Local Differential Privacy
• Bit-Plane Slicing
• Privacy-Preserving Machine Learning
• Perceptual Obfuscation
• Privacy Budget Allocation