Adaptive Gaussian Mechanism Based on Expected Data Utility under Conditional Filtering Noise

doi:10.3837/tiis.2018.07.027

Cited by 3 publications

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

DPBA-WGAN: A Vector-Valued Differential Private Bilateral Alternative Scheme on WGAN for Image Generation

Zhang

2023

IEEE Access

View full text Add to dashboard Cite

The large amount of sensitive personal information used in deep learning models has attracted considerable attention for privacy security. Sensitive data may be memorialized or encoded into the parameters or generation of the Wasserstein Generative Adversarial Networks (WGAN), which can be prevented by implementing privacy-preserving algorithms during the parameter training process. Meanwhile, the model is also expected to obtain effective generated results. We propose a vector-valued differential private bilateral alternative (DPBA) algorithm, a novel perturbation method for the training process. The vector-valued Gaussian (VVG) noise involving functional structure information is injected into the WGAN to generate data with privacy protection, and the model is verified to satisfy differential privacy. The bilateral alternative noise can eventually randomly perturb the gradient and generates informative feature-rich samples. The dynamic noise and vector-based perturbation approach ensures privacy strength. After extensive evaluation, our algorithm outperformed state-of-the-art techniques in terms of usability metrics for all validation datasets. The downstream classification accuracy for the generated Modified National Institute of Standards and Technology (Mnist) was 97.04%, whereas that for the Fashion Modified National Institute of Standards and Technology (Fashion-Mnist) dataset was 80.91%. Mnist improved the average accuracy of the neural network classifier by at least 16.81%, and Fashion-Mnist by at least 3.55%. In the multichannel generation tasks, the binary classification accuracy improved by at least 10.4% compared to CelebFaces Attributes (CelebA), and the accuracy of the Street View House Numbers (SVHN) was as high as 86.1%. The perturbation method proved highly resilient to gradient attack recovery under simulated gradient attacks.

show abstract

DPBA-WGAN: A Vector-Valued Differential Private Bilateral Alternative Scheme on WGAN for Image Generation

Zhang

2023

IEEE Access

View full text Add to dashboard Cite

show abstract

Balancing Privacy-Utility of Differential Privacy Mechanism: A Collaborative Perspective

Liu

Peng

Tian

et al. 2021

Security and Communication Networks

View full text Add to dashboard Cite

Differential privacy mechanism can maintain privacy-utility monotonicity. Thus, differential privacy mechanism does not obtain privacy-utility balance for numerical data. To this end, we provide privacy-utility balance of differential privacy mechanism with the collaborative perspective in this paper. First, we constructed the collaborative model achieving privacy-utility balance of differential privacy mechanism. Second, we presented the collaborative algorithm of differential privacy mechanism under our collaborative model. Third, our theoretical analysis showed that the collaborative algorithm of differential privacy mechanism could keep privacy-utility balance. Finally, our experimental results demonstrated that the collaborative differential privacy mechanism can maintain privacy-utility balance. Thus, we provide a new collaborative model to solve the privacy-utility balance problem of differential privacy mechanism. Our collaborative algorithm is easy to apply to query processing of numerical data.

show abstract

Reviewing and Improving the Gaussian Mechanism for Differential Privacy

Zhao¹,

Wang²,

Bai³

et al. 2019

Preprint

View full text Add to dashboard Cite

Differential privacy provides a rigorous framework to quantify data privacy, and has received considerable interest recently. A randomized mechanism satisfying (ǫ, δ)-differential privacy (DP) roughly means that, except with a small probability δ, altering a record in a dataset cannot change the probability that an output is seen by more than a multiplicative factor e ǫ . A well-known solution to (ǫ, δ)-DP is the Gaussian mechanism initiated by Dwork et al. [1] in 2006 with an improvement by Dwork and Roth [2] in 2014, where a Gaussian noise amount 2 ln 2 δ × ∆ ǫ of [1] or 2 ln 1.25 δ × ∆ ǫ of [2] is added independently to each dimension of the query result, for a query with ℓ2-sensitivity ∆. Although both classical Gaussian mechanisms [1], [2] explicitly assume 0 < ǫ ≤ 1 only, our review finds that many studies in the literature have used the classical Gaussian mechanisms under values of ǫ and δ where we show the added noise amounts of [1], [2] do not achieve (ǫ, δ)-DP. We obtain such result by analyzing the optimal (i.e., least) Gaussian noise amount σDP-OPT for (ǫ, δ)-DP and identifying the set of ǫ and δ where the noise amounts of classical Gaussian mechanisms are even less than σDP-OPT. The inapplicability of mechanisms of [1],[2] to large ǫ can also be seen from our result that σDP-OPT for large ǫ can be written as Θ 1 √ ǫ , but not Θ 1 ǫ . Since σDP-OPT has no closed-form expression and needs to be approximated in an iterative manner, we propose Gaussian mechanisms by deriving closed-form upper bounds for σDP-OPT. Our mechanisms achieve (ǫ, δ)-DP for any ǫ, while the classical Gaussian mechanisms [1], [2] do not achieve (ǫ, δ)-DP for large ǫ given δ. Moreover, the utilities of our proposed Gaussian mechanisms improve those of the classical Gaussian mechanisms [1],[2] and are close to that of the optimal yet more computationally expensive Gaussian mechanism.Since most mechanisms proposed in the literature for (ǫ, δ)-DP are obtained by ensuring a condition called (ǫ, δ)-probabilistic differential privacy (pDP), we also present an extensive discussion of (ǫ, δ)-pDP including deriving Gaussian noise amounts to achieve it.To summarize, our paper fixes the literature's long-time misuse of Gaussian mechanism [1], [2] for (ǫ, δ)-differential privacy and provides a comprehensive study for the Gaussian mechanisms.

show abstract

Adaptive Gaussian Mechanism Based on Expected Data Utility under Conditional Filtering Noise

Cited by 3 publications

References 20 publications

DPBA-WGAN: A Vector-Valued Differential Private Bilateral Alternative Scheme on WGAN for Image Generation

DPBA-WGAN: A Vector-Valued Differential Private Bilateral Alternative Scheme on WGAN for Image Generation

Balancing Privacy-Utility of Differential Privacy Mechanism: A Collaborative Perspective

Reviewing and Improving the Gaussian Mechanism for Differential Privacy

Contact Info

Product

Resources

About