The Algorithmic Foundations of Differential Privacy

By Cynthia Dwork, Aaron Roth. 2014

9,629 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Privacy & Security

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Privacy & Security

Citations in the corpus

Privacy-Preserving Generative Deep Neural Networks Support Clinical Data Sharing (2017, background): “Differential privacy protects against common privacy attacks including membership inference, homogeneity and background knowledge attacks (23).”
Towards the Science of Security and Privacy in Machine Learning (2016, background): “This is captured by differential privacy [95], a rigorous framework to analyze the privacy guarantees provided by algorithms.”
Deep Learning with Differential Privacy (2016, background): “In a concurrent and independent work Bun and Steinke [11] introduce a relaxation of differential privacy (generalizing the work of Dwork and Rothblum [22]) defined via a linear upper”
Advances and Open Problems in Federated Learning (2019, background): “It is stronger than the typically used notion of adjacency where D and D′ differ by only one record [145], since in general one user may contribute many records (e.”
Federated Learning With Differential Privacy: Algorithms and Performance Analysis (2019, methodology): “For numerical data, a Gaussian mechanism defined in [24] can be used to guarantee ( , δ)-DP.”
Deep Models Under the GAN: Information Leakage from Collaborative Deep Learning (2017, methodology): “Differential privacy for big data was considered in [22].”
Semi-supervised Knowledge Transfer for Deep Learning from Private Training Data (2016, methodology): “However, its lack of randomization gives rise to caveats (Dwork & Roth, 2014), and attackers can infer properties of the dataset (Aggarwal, 2005).”
Certified Robustness to Adversarial Examples with Differential Privacy (2018, background): “, when ‖α‖p ≤ L, by applying group privacy [17] (see §III-B).”
Practical and Private (Deep) Learning without Sampling or Shuffling (2021, background): “25 log(2 /δ ) /σ, δ -differential privacy [19, 21] and ( α, α/ 2 σ where clip ( v, L ) projects v to the (cid:96) 2 -ball of radius L , and B ⊆ [ n ] represents a mini-batch of data.”
DISCO: Dynamic and Invariant Sensitive Channel Obfuscation for deep neural networks (2020, methodology): “Additionally, techniques such as multi-party computation [14, 15] and differential privacy [16, 17, 18, 19] have been employed to improve the privacy in federated-learning.”
Training Production Language Models without Memorizing User Data (2020, background): “However, extending to the case of n u = 16 users (e.g., via a group privacy argument [DR + 14]) will result in a very weak protection.”
Tempered Sigmoid Activations for Deep Learning with Differential Privacy (2020, background): “To counter this, ML techniques that offer strong guarantees expressed in the framework of differential privacy (Dwork and Roth 2014) have been developed.”
Learning discrete distributions: user vs item-level privacy (2020, background): “Diﬀerential privacy [Dwork et al., 2006, Dwork and Roth, 2014, Wasserman and Zhou, 2010] has emerged as the standard framework for providing privacy for various statistical problems.”
Privacy Amplification via Random Check-Ins (2020, methodology): “2 and applying advanced composition for DP [17].”
Federated Learning and Differential Privacy: Software tools analysis, the Sherpa.ai FL framework and methodological guidelines for preserving data privacy (2020, methodology): “Nevertheless, the resulting distribution can be rather complex and over an arbitrarily large domain, thus the implementation of such mechanism might not always be efﬁcient [12].”