Hypothesis Testing Approach to Differential Privacy with an information Theoretic Perspective
Event details
| Date | 09.06.2023 |
| Hour | 10:00 › 12:00 |
| Speaker | Cemre Cadir |
| Location | |
| Category | Conferences - Seminars |
EDIC candidacy exam
Exam president: Prof. Emre Telatar
Thesis advisor: Prof. Michael Gastpar
Thesis co-advisor: Prof. Yanina Shkel
Co-examiner: Prof. Patrick Thiran
Abstract
In the scope of this write-up three papers offering
a hypothesis testing approach to differential privacy (DP) are
studied in depth. First, the operational interpretation of DP
based on hypothesis testing is investigated and the resulting
advanced composition theorem is explained [1]. Second, fdifferential
privacy (f-DP) is defined and its properties are
discussed including a central limit theorem like phenomena under
composition [2]. Finally, the relationship of Renyi differential
privacy (RDP) with DP and f-DP is inspected and the improved
privacy analysis method is examined.
Background papers
1. "Gaussian Differential Privacy", in Journal of the Royal Statistical Society Series B: Statistical Methodology, Volume 84, Issue 1, February 2022, Pages 3–37, https://doi.org/10.1111/rssb.12454 (Excluding the supplemental materials, appendixes and discussions) https://academic.oup.com/jrsssb/article/84/1/3/7056089?login=false, by Jinshuo Dong, Aaron Roth, Weijie J. S.
2. "Three Variants of Differential Privacy: Lossless Conversion and Applications," in IEEE Journal on Selected Areas in Information Theory, vol. 2, no. 1, pp. 208-222, March 2021, doi: 10.1109/JSAIT.2021.3054692. https://ieeexplore.ieee.org/document/9336023 by S. Asoodeh, J. Liao, F. P. Calmon, O. Kosut and L. Sankar.
3. "The Composition Theorem for Differential Privacy," in IEEE Transactions on Information Theory, vol. 63, no. 6, pp. 4037-4049, June 2017, doi: 10.1109/TIT.2017.2685505. https://ieeexplore.ieee.org/document/7883827, by P. Kairouz, S. Oh and P. Viswanath.
Exam president: Prof. Emre Telatar
Thesis advisor: Prof. Michael Gastpar
Thesis co-advisor: Prof. Yanina Shkel
Co-examiner: Prof. Patrick Thiran
Abstract
In the scope of this write-up three papers offering
a hypothesis testing approach to differential privacy (DP) are
studied in depth. First, the operational interpretation of DP
based on hypothesis testing is investigated and the resulting
advanced composition theorem is explained [1]. Second, fdifferential
privacy (f-DP) is defined and its properties are
discussed including a central limit theorem like phenomena under
composition [2]. Finally, the relationship of Renyi differential
privacy (RDP) with DP and f-DP is inspected and the improved
privacy analysis method is examined.
Background papers
1. "Gaussian Differential Privacy", in Journal of the Royal Statistical Society Series B: Statistical Methodology, Volume 84, Issue 1, February 2022, Pages 3–37, https://doi.org/10.1111/rssb.12454 (Excluding the supplemental materials, appendixes and discussions) https://academic.oup.com/jrsssb/article/84/1/3/7056089?login=false, by Jinshuo Dong, Aaron Roth, Weijie J. S.
2. "Three Variants of Differential Privacy: Lossless Conversion and Applications," in IEEE Journal on Selected Areas in Information Theory, vol. 2, no. 1, pp. 208-222, March 2021, doi: 10.1109/JSAIT.2021.3054692. https://ieeexplore.ieee.org/document/9336023 by S. Asoodeh, J. Liao, F. P. Calmon, O. Kosut and L. Sankar.
3. "The Composition Theorem for Differential Privacy," in IEEE Transactions on Information Theory, vol. 63, no. 6, pp. 4037-4049, June 2017, doi: 10.1109/TIT.2017.2685505. https://ieeexplore.ieee.org/document/7883827, by P. Kairouz, S. Oh and P. Viswanath.
Practical information
- General public
- Free