Distribution-Invariant Differential Privacy
Preprint
Abstract
Differential privacy is becoming one gold standard for protecting the privacy of publicly shared data. It has been widely used in social science, data science, public health, information technology, and the U.S. decennial census. Nevertheless, to guarantee differential privacy, existing methods may unavoidably alter the conclusion of the original data analysis, as privatization often changes the sample distribution. This phenomenon is known as the trade-off between privacy protection and statistical accuracy. In this work, we mitigate this trade-off by developing a distribution-invariant privatization (DIP) method to reconcile both high statistical accuracy and strict differential privacy. As a result, any downstream statistical or machine learning task yields essentially the same conclusion as if one used the original data. Numerically, under the same strictness of privacy protection, DIP achieves superior statistical accuracy in a wide range of simulation studies and real-world benchmarks.Keywords
All Related Versions
This publication has 27 references indexed in Scilit:
- Forecasting influenza activity using machine-learned mobility mapNature Communications, 2021
- AI reflections in 2020Nature Machine Intelligence, 2021
- Privacy-Preserving Parametric Inference: A Case for Robust StatisticsJournal of the American Statistical Association, 2020
- Transparency and reproducibility in artificial intelligenceNature, 2020
- Secure, privacy-preserving and federated machine learning in medical imagingNature Machine Intelligence, 2020
- How differential privacy will affect our understanding of health disparities in the United StatesProceedings of the National Academy of Sciences, 2020
- Towards formalizing the GDPR’s notion of singling outProceedings of the National Academy of Sciences, 2020
- Minimax Optimal Procedures for Locally Private EstimationJournal of the American Statistical Association, 2018
- Private algorithms for the protected in social network searchProceedings of the National Academy of Sciences, 2016
- A data-driven approach to predict the success of bank telemarketingDecision Support Systems, 2014