Learning Differentially Private Mechanisms
This work provides automated support for correctly constructing differentially private algorithms, which is a non-trivial task prone to errors, benefiting researchers and practitioners in data privacy.
This paper addresses the challenge of automatically converting non-private programs into differentially private versions. The authors propose a technique that combines representative example inputs, continuous optimization, and symbolic expression mapping, demonstrating its ability to learn foundational differential privacy algorithms and outperform program synthesis baselines.
Differential privacy is a formal, mathematical definition of data privacy that has gained traction in academia, industry, and government. The task of correctly constructing differentially private algorithms is non-trivial, and mistakes have been made in foundational algorithms. Currently, there is no automated support for converting an existing, non-private program into a differentially private version. In this paper, we propose a technique for automatically learning an accurate and differentially private version of a given non-private program. We show how to solve this difficult program synthesis problem via a combination of techniques: carefully picking representative example inputs, reducing the problem to continuous optimization, and mapping the results back to symbolic expressions. We demonstrate that our approach is able to learn foundational algorithms from the differential privacy literature and significantly outperforms natural program synthesis baselines.