Filters








21,324 Hits in 5.1 sec

Making the Most of Parallel Composition in Differential Privacy [article]

Josh Smith and Hassan Jameel Asghar and Gianpaolo Gioiosa and Sirine Mrabet and Serge Gaspers and Paul Tyler
2021 arXiv   pre-print
We prove the parallel composition theorem for f-differential privacy. We evaluate our approach on synthetic and real-world data sets of queries.  ...  Our approach is defined in the general setting of f-differential privacy, which subsumes standard pure differential privacy and Gaussian differential privacy.  ...  This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.  ... 
arXiv:2109.09078v1 fatcat:njb25rzwh5dp3ou6zpeuvczlmi

Making the Most of Parallel Composition in Differential Privacy

Josh Smith, Hassan Jameel Asghar, Gianpaolo Gioiosa, Sirine Mrabet, Serge Gaspers, Paul Tyler
2021 Proceedings on Privacy Enhancing Technologies  
We prove the parallel composition theorem for f-differential privacy. We evaluate our approach on synthetic and real-world data sets of queries.  ...  Our approach is defined in the general setting of f-differential privacy, which subsumes standard pure differential privacy and Gaussian differential privacy.  ...  Our approach addresses the problem of making optimal use of parallel composition directly, thereby avoiding the significant computational overheads associated with most of the optimisation routines used  ... 
doi:10.2478/popets-2022-0013 fatcat:dvjhrf7pyjfafkansppry2z44u

Differential identifiability clustering algorithms for big data analysis

Tao Shang, Zheng Zhao, Xujie Ren, Jianwei Liu
2021 Science China Information Sciences  
The definition of ρ-differential identifiability (DI) precisely matches the legal definitions of privacy, which can provide an easy parameterization approach for practitioners so that they can set privacy  ...  In this paper, we propose a new exponential mechanism and composition properties of differential identifiability, and then apply differential identifiability to k-means and k-prototypes algorithms on MapReduce  ...  We also develop sequential composition and parallel composition of differential identifiability for complex privacy preservation issues.  ... 
doi:10.1007/s11432-020-2910-1 fatcat:4odlynwnejapjoxgebuhwmrrea

Differentially Private Algorithms for 2020 Census Detailed DHC Race & Ethnicity [article]

Sam Haney and William Sexton and Ashwin Machanavajjhala and Michael Hay and Gerome Miklau
2021 arXiv   pre-print
We analytically estimate the privacy loss parameters ensured by the two algorithms for comparable levels of error introduced in the statistics.  ...  Discrete Gaussian distribution that satisfied a well studied variant of differential privacy, called Zero Concentrated Differential Privacy (zCDP).  ...  There are two types of composition we are interested in -sequential composition and parallel composition. We first state the sequential composition results. Lemma 1.  ... 
arXiv:2107.10659v1 fatcat:re6mrj566zhixdhan6rb33n74q

Sampling and partitioning for differential privacy

Hamid Ebadi, Thibaud Antignac, David Sands
2016 2016 14th Annual Conference on Privacy, Security and Trust (PST)  
The most promising among them completely discharge the user of the privacy concerns by transparently taking care of the privacy budget.  ...  of sampling methods and show how they can be correctly implemented in a system for differential privacy.  ...  Sharing of a privacy budget between two analyses is an important outcome of parallel composition.  ... 
doi:10.1109/pst.2016.7906954 dblp:conf/pst/EbadiAS16 fatcat:jcwfutftjbcghler36orsn3b7u

Differential Privacy

Hamid Ebadi, David Sands, Gerardo Schneider
2015 Proceedings of the 42nd Annual ACM SIGPLAN-SIGACT Symposium on Principles of Programming Languages - POPL '15  
It is based on a simple generalisation of classic differential privacy which we call Personalised Differential Privacy (PDP). In PDP each individual has its own personal privacy level.  ...  Alice determines the multiplier for the privacy decrease in Alice's budget.  ...  Many thanks to our colleagues in the ProSec and Formal Methods groups for many helpful discussions, and special thanks Raúl Pardo Jiménez for participation in the early stage of the research, and to Niklas  ... 
doi:10.1145/2676726.2677005 dblp:conf/popl/EbadiSS15 fatcat:fhngdnmekrftnixu4dljfm4rw4

Applications of Differential Privacy in Social Network Analysis: A Survey [article]

Honglu Jiang, Jian Pei, Dongxiao Yu, Jiguo Yu, Bei Gong, Xiuzhen Cheng
2021 arXiv   pre-print
We present a concise review of the foundations of differential privacy and the major variants and discuss how differential privacy is applied to social network analysis, including privacy attacks in social  ...  In this article, we provide a comprehensive survey connecting the basic principles of differential privacy and applications in social network analysis.  ...  The forms of privacy guarantee are the same. Therefore, local differential privacy inherits the sequential and parallel composition features mentioned in Section 2.4.  ... 
arXiv:2010.02973v2 fatcat:finuu7fjenhzzcfhplapwt37mi

More Than Privacy: Applying Differential Privacy in Key Areas of Artificial Intelligence [article]

Tianqing Zhu and Dayong Ye and Wei Wang and Wanlei Zhou and Philip S. Yu
2020 arXiv   pre-print
For this reason, differential privacy has been broadly applied in AI but to date, no study has documented which differential privacy mechanisms can or have been leveraged to overcome its issues or the  ...  It can also be used to improve security, stabilize learning, build fair models, and impose composition in selected areas of AI.  ...  Composition Two privacy budget composition theorems are widely used in the design of differential privacy mechanisms: sequential composition [14] and parallel composition [15] .  ... 
arXiv:2008.01916v1 fatcat:ujmxv7eq6jcppndfu5shbzkdom

More Than Privacy: Applying Differential Privacy in Key Areas of Artificial Intelligence

Tianqing Zhu, Dayong Ye, Wei Wang, Wanlei Zhou, Philip Yu
2020 IEEE Transactions on Knowledge and Data Engineering  
For this reason, differential privacy has been broadly applied in AI but to date, no study has documented which differential privacy mechanisms can or have been leveraged to overcome its issues or the  ...  It can also be used to improve security, stabilize learning, build fair models, and impose composition in selected areas of AI.  ...  In this summary, we can see that most of these papers make use of the Gaussian mechanism.  ... 
doi:10.1109/tkde.2020.3014246 fatcat:33rl6jxy5rgexpnuel5rvlkg5a

Information-Theoretic Bounds for Differentially Private Mechanisms

Gilles Barthe, Boris Kopf
2011 2011 IEEE 24th Computer Security Foundations Symposium  
Each line of research has developed its own notion of confidentiality: on the one hand, differential privacy is the emerging consensus guarantee used for privacy-preserving data analysis.  ...  Finally, we show that the question of providing optimal upper bounds for the leakage of -differentially private mechanisms in terms of rational functions of is in fact decidable.  ...  Acknowledgments The authors would like to thank Miguel Andrés, Catuscia Palamidessi, and the anonymous reviewers for their helpful feedback. This research was supported by FP7-ICT Project NESSoS  ... 
doi:10.1109/csf.2011.20 dblp:conf/csfw/BartheK11 fatcat:fvgyg6bvvffrhbauh23bflgmsy

The Philosophy of Differential Privacy

Claire McKay Bowen, Simson Garfinkel
2021 Notices of the American Mathematical Society  
Most differentially private methods try to leverage parallel composition instead of sequential to avoid splitting the privacy loss budget, because smaller means less accuracy.  ...  (a) Sequential composition. The sequence of ℳ ( ) applied on the same provides (∑ , ∑ )differential privacy. (b) Parallel composition. Let be disjoint subsets of the input domain .  ... 
doi:10.1090/noti2363 fatcat:pe2ri2ukynf7hhkeonvnwxhpvq

Improving the Utility of Differential Privacy via Univariate Microaggregation [chapter]

David Sánchez, Josep Domingo-Ferrer, Sergio Martínez
2014 Lecture Notes in Computer Science  
The theoretical benefits of the proposal are illustrated and in a practical setting.  ...  Our proposal relies on univariate microaggregation to reduce the amount of noise needed to satisfy differential privacy.  ...  Lemma 3 (Parallel composition [25] ). Let each sanitizing algorithm Ag i in a set of sanitizers provide ε-differential privacy.  ... 
doi:10.1007/978-3-319-11257-2_11 fatcat:f22ifjgrqnfvpfdynm4w6usg24

Modular Reasoning about Differential Privacy in a Probabilistic Process Calculus [chapter]

Lili Xu
2013 Lecture Notes in Computer Science  
Finally, we make some preliminary steps towards automatically computing the degree of privacy of a system in a compositional way.  ...  In addition to the standard parallel composition and nondeterministic choice of CCS, CCS p provides also a primitive for the probabilistic choice.  ...  The compositionality results in our paper are closely related to those of [5] , although we use a different measure of protection (differential privacy). -Compositionality of Differential Privacy.  ... 
doi:10.1007/978-3-642-41157-1_13 fatcat:uicpciq5tzcsjaw2ss3jj2fs2e

Plume: Differential Privacy at Scale [article]

Kareem Amin and Jennifer Gillenwater and Matthew Joseph and Alex Kulesza and Sergei Vassilvitskii
2022 arXiv   pre-print
Differential privacy has become the standard for private data analysis, and an extensive literature now offers differentially private solutions to a wide variety of problems.  ...  We describe a number of sometimes subtle implementation issues and offer practical solutions that, together, make an industrial-scale system for differentially private data analysis possible.  ...  Acknowledgments We would like to thank Per Anderson, Christoph Dibak, Miguel Guevara, Andrés Muñoz Medina, and Jane Shapiro for their critical work in making Plume possible.  ... 
arXiv:2201.11603v1 fatcat:h26yrcr6o5dkrakb5ahdqg5dce

Differentially Private Publication of Social Graphs at Linear Cost

Hiep H. Nguyen, Abdessamad Imine, Michaël Rusinowitch
2015 Proceedings of the 2015 IEEE/ACM International Conference on Advances in Social Networks Analysis and Mining 2015 - ASONAM '15  
The problem of private publication of graph data has attracted a lot of attention recently. The prevalence of differential privacy makes the problem more promising.  ...  However, a large body of existing works on differentially private release of graphs have not answered the question about the upper bounds of privacy budgets.  ...  Theorem 2.2: (Sequential and parallel compositions [9] ) Let each A i provide ǫ i -differential privacy. A sequence of A i (D) over the dataset D provides Σ n i=1 ǫ i -differential privacy.  ... 
doi:10.1145/2808797.2809385 dblp:conf/asunam/NguyenIR15 fatcat:gsvy7cnrpfcrhd74felvpgeccm
« Previous Showing results 1 — 15 out of 21,324 results