Certification for Differentially Private Prediction in Gradient-Based Training

TL;DR

This paper presents a new method for differentially private prediction that uses dataset-specific sensitivity bounds, leading to better privacy-utility trade-offs than traditional global sensitivity approaches.

Contribution

We introduce a novel technique combining convex relaxation and bound propagation to compute dataset-specific sensitivity bounds for private prediction.

Findings

Sensitivity bounds are significantly tighter than global sensitivity.

Experimental results show improved privacy-utility trade-offs.

Method applied successfully to medical imaging and NLP datasets.

Abstract

We study private prediction where differential privacy is achieved by adding noise to the outputs of a non-private model. Existing methods rely on noise proportional to the global sensitivity of the model, often resulting in sub-optimal privacy-utility trade-offs compared to private training. We introduce a novel approach for computing dataset-specific upper bounds on prediction sensitivity by leveraging convex relaxation and bound propagation techniques. By combining these bounds with the smooth sensitivity mechanism, we significantly improve the privacy analysis of private prediction compared to global sensitivity-based approaches. Experimental results across real-world datasets in medical image classification and natural language processing demonstrate that our sensitivity bounds are can be orders of magnitude tighter than global sensitivity. Our approach provides a strong basis for the development of novel privacy preserving technologies.