Geometry of Sensitivity: Twice Sampling and Hybrid Clipping in Differential Privacy with Optimal Gaussian Noise and Application to Deep Learning
We study the fundamental problem of the construction of optimal randomization in Differential Privacy (DP). Depending on the clipping strategy or additional properties of the processing function, the corresponding sensitivity set theoretically determines the necessary randomization to produce the re...
Päätekijät: | Xiao, Hanshen, Wan, Jun, Devadas, Srinivas |
---|---|
Muut tekijät: | Massachusetts Institute of Technology. Computer Science and Artificial Intelligence Laboratory |
Aineistotyyppi: | Artikkeli |
Kieli: | English |
Julkaistu: |
ACM|Proceedings of the 2023 ACM SIGSAC Conference on Computer and Communications Security
2023
|
Linkit: | https://hdl.handle.net/1721.1/153139 |
Samankaltaisia teoksia
-
Deep Learning with Gaussian Differential Privacy
Tekijä: Zhiqi Bu, et al.
Julkaistu: (2020-07-01) -
Formal Privacy Proof of Data Encoding: The Possibility and Impossibility of Learnable Encryption
Tekijä: Xiao, Hanshen, et al.
Julkaistu: (2025) -
Local differential privacy in decentralized optimization
Tekijä: Xiao, Hanshen.
Julkaistu: (2020) -
Enhancing Efficiency in Privacy-Preserving Federated Learning for Healthcare: Adaptive Gaussian Clipping With DFT Aggregator
Tekijä: Muhammad Ayat Hidayat, et al.
Julkaistu: (2024-01-01) -
Adaptive selection method of differential privacy GAN gradient clipping thresholds
Tekijä: Peng GUO, et al.
Julkaistu: (2018-05-01)