Foundations of Differential Privacy

Principal Investigator: Jordan Awan

Differential privacy (DP) has arisen as the state-of-the-art framework for formal privacy protection when analyzing sensitive data. However, the fundamentals of DP are still not well understood. There are many important questions for DP, which with proper understanding can help researchers understand the fundamental limits of what can be achieved with privacy protection as well as the optimal methods to achieve DP. In this project, we study the foundations of DP such as 1) the choice of privacy definition, 2) optimal noise to achieve a given DP definition, and 3) understanding the sensitivity of a statistic and how this relates to how much noise is required to protect privacy.

Personnel

Other Faculty: Aleksandra Slavkovic, Professor of Statistics, Penn State Salil Vadhan, Vicky Joseph Professor of Computer Science and Applied Mathematics, Harvard University

Students: Aishwarya Ramasethu Young Hyun Cho

Representative Publications

• Awan, Jordan, and Aishwarya Ramasethu. "Optimizing Noise for $ f $-Differential Privacy via Anti-Concentration and Stochastic Dominance." arXiv preprint arXiv:2308.08343 (2023).
   
  Awan, Jordan, and Salil Vadhan. "Canonical noise distributions and private hypothesis tests." The Annals of Statistics 51, no. 2 (2023): 547-572.

   

  Awan, Jordan, and Jinshuo Dong. "Log-concave and multivariate canonical noise distributions for differential privacy." Advances in Neural Information Processing Systems 35 (2022): 34229-34240.

   

  Awan, Jordan, and Aleksandra Slavković. "Structure and sensitivity in differential privacy: Comparing k-norm mechanisms." Journal of the American Statistical Association 116, no. 534 (2021): 935-954.

   

   

Keywords: Differential Privacy, sensitivity