Differential Privacy

By Cynthia Dwork. 2006

7,564 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Sociology and Political Science, Safety, Robustness & Interpretability

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 Sociology and Political Science

In Safety, Robustness & Interpretability

Citations in the corpus

Privacy-Preserving Generative Deep Neural Networks Support Clinical Data Sharing (2017, background): “Differential privacy is a stability property for algorithms, specifically for randomized algorithms [37].”
Membership Inference Attacks Against Machine Learning Models (2016, background): “If the training process is differentially private [12], the probability of producing a given model from a training dataset that includes a particular record is close to the probability of producing the same model when this record is not included.”
Calibrating Noise to Sensitivity in Private Data Analysis (2006, methodology): “This paper is an update of the conference version Dwork et al. (2006b), with revisions to presentation and terminology. We do not aim to survey all the subsequent work on differential privacy. At the end of each section, we briefly discuss how the technical ideas we present have affected subsequent ”
Model Inversion Attacks that Exploit Confidence Information and Basic Countermeasures (2015, background): “The framework gives rise to a number of security notions, including variants of differential privacy [11].”
Deep Models Under the GAN: Information Leakage from Collaborative Deep Learning (2017, background): “lgorithms trained on trusted processors. Differential privacy plays an important role in deep learning as shown in [1, 39, 77, 79]. 4.3 Differential Privacy DifferentialPrivacy(DP)wasintroducedbyDwork[21].Itsaimisto provide provable privacy guarantees for database records without significant query a”
Backdoor attacks and defenses in feature-partitioned collaborative learning (2020, background): “Privacy-preserving techniques such as Homomorphic Encryption (HE) (Rivest et al., 1978; Acar et al., 2018), Secure Multi-party Computation (SMPC) (Yao, 1982) and Differential Privacy (DP) (Dwork, 2006) are typically applied in these frameworks to preserve user privacy and data confidentiality.”
Rényi Differential Privacy of the Sampled Gaussian Mechanism (2019, background): “The notion of differential privacy [Dwo06, DMNS06] grounds a guarantee of individual privacy for input of a statistical procedure in non-determinism of the procedure’s output.”
Preserving ON-OFF Privacy for Past and Future Requests (2019, methodology): “Several existing privacy-preserving solutions can be utilized to ensure a desired level of privacy for the user, such as anonymity [1], differential privacy [2], private information retrieval [3], to name a few.”
Differential Privacy Has Disparate Impact on Model Accuracy (2019, background): “-differential privacy (DP) [12] bounds the inﬂuence of any single input on the output of a computation.”
ON-OFF Privacy with Correlated Requests (2019, background): “This is now a well-acknowledged concern and has lead to many interesting theoretical problems such as anonymity [1], differential privacy [2], private information retrieval [3], and other privacy-preserving algorithms.”
Data Poisoning against Differentially-Private Learners: Attacks and Defenses (2019, methodology): “In this paper, we study differential privacy [8, 7] as a general defensive technique against data poisoning.”
Fully Decentralized Joint Learning of Personalized Models and Collaboration Graphs (2019, background): “Another promising direction is to make our approach differentially-private (Dwork, 2006) to formally guarantee that personal datasets cannot be inferred from the information sent by users.”
APPLIED FEDERATED LEARNING: IMPROVING GOOGLE KEYBOARD QUERY SUGGESTIONS (2018, methodology): “Language models have also been trained with FL and differential privacy for further privacy enhancements [9, 10].”
Amplification by Shuffling: From Local to Central Differential Privacy via Anonymity (2018, background): “…work has proposed a number of mechanisms that provide users with strong privacy-protection guarantees in terms of of differential privacy [DMNS06, Dwo06, KLN + 08] and, speciﬁcally, mechanisms that provide local differential privacy (LDP) have been deployed by Google, Apple, and Microsoft…”
Federated Learning for Mobile Keyboard Prediction (2018, methodology): “Federated learning has previously been shown to be complementary to privacy-preserving techniques such as secure aggregation [26] and differential privacy [17].”