Graphical-model based estimation and inference for differential privacy
This work addresses efficiency and accuracy challenges in differential privacy mechanisms for data analysis, representing an incremental improvement over existing methods.
The paper tackles the problem of efficiently estimating answers to new queries from noisy privacy measurements, particularly in high-dimensional distributions with low-dimensional marginals, and shows that their graphical model approach is far more efficient than existing techniques and improves accuracy and scalability.
Many privacy mechanisms reveal high-level information about a data distribution through noisy measurements. It is common to use this information to estimate the answers to new queries. In this work, we provide an approach to solve this estimation problem efficiently using graphical models, which is particularly effective when the distribution is high-dimensional but the measurements are over low-dimensional marginals. We show that our approach is far more efficient than existing estimation techniques from the privacy literature and that it can improve the accuracy and scalability of many state-of-the-art mechanisms.