A Differential Privacy Mechanism Combining Randomized Response with Laplace Distribution
Youqin Chen, Zhengquan Xu, Jian‐Zhang Chen · 2021 7th International Conference on Computer and Communications (ICCC) · 2021
By combining randomized response with Laplace distribution, it obtains a new kind of differential privacy mechanism called Pseudo-Laplace mechanism. By analyzing the privacy and data availability of the mechanism, as well as an optimization problem that provides better privacy protection than the Laplace mechanism based solely on the Laplace distribution (traditional Laplace mechanism) under the same data availability, and the instantiations of the mechanism obtained in the approximate optimization algorithm, it finds that the mechanism could have almost the same privacy protection level in the middle and high privacy protection region (approximately ε < 5 ) and better privacy protection level in the low privacy protection region (approximately ε ≥5 ) than the traditional Laplace mechanism under the same data availability. It has important practical implications for many practical applications where data availability is prioritized while better user privacy is protected.