Search papers, labs, and topics across Lattice.
The paper introduces LDP-Slicing, a novel framework for applying Local Differential Privacy (LDP) to images by decomposing pixel values into binary bit-planes, addressing the utility degradation issues of canonical LDP mechanisms in high-dimensional pixel spaces. LDP-Slicing incorporates perceptual obfuscation and optimization-based privacy budget allocation to enhance privacy and maintain utility. Experiments on face recognition and image classification show that LDP-Slicing achieves superior performance compared to existing DP/LDP baselines with minimal computational overhead.
Applying local differential privacy to images doesn't have to destroy utility: decompose pixels into bit-planes and suddenly LDP becomes practical for downstream tasks.
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.
