37,097 Hits in 3.9 sec

Mutual Information Optimally Local Private Discrete Distribution Estimation [article]

Shaowei Wang, Liusheng Huang, Pengzhan Wang, Yiwen Nie, Hongli Xu, Wei Yang, Xiang-Yang Li, Chunming Qiao
2016 arXiv   pre-print
Consider statistical learning (e.g. discrete distribution estimation) with local ϵ-differential privacy, which preserves each data provider's privacy locally, we aim to optimize statistical data utility  ...  After analysing the limitations of existing local private mechanisms from mutual information perspective, we propose an efficient implementation of the k-subset mechanism for discrete distribution estimation  ...  In this work, we study local private data utilities for full privacy region, mainly focus on mutual information and discrete distribution estimation.  ... 
arXiv:1607.08025v1 fatcat:6euqmewcdrcrjgn6uz5zw36lmy

Towards Distributed Privacy-Preserving Prediction [article]

Lingjuan Lyu, Yee Wei Law, Kee Siong Ng, Shibei Xue, Jun Zhao, Mengmeng Yang, Lei Liu
2020 arXiv   pre-print
First, we introduce the improved Binomial Mechanism and Discrete Gaussian Mechanism to achieve distributed differential privacy.  ...  Experimental results demonstrate that our framework has comparable performance to the non-private frameworks and delivers better results than the local differentially private framework and standalone framework  ...  train a global model; (2) Distributed non-private framework excludes both DP and cryptosystem, teachers directly share their local predictions with the aggregator; (3) Local differentially private (LDP  ... 
arXiv:1910.11478v2 fatcat:xa3m7wzruzfmlhgqlikdop67im

Locally Differentially Private Naive Bayes Classification [article]

Emre Yilmaz, Mohammad Al-Rubaie, J. Morris Chang
2019 arXiv   pre-print
We propose solutions for both discrete and continuous data.  ...  The data aggregator estimates all probabilities needed by the Naive Bayes classifier. Then, new instances can be classified based on the estimated probabilities.  ...  In local differential privacy (LDP), individuals send their data to the data aggregator after privatizing data by perturbation. Hence, these techniques provide plausible deniability for individuals.  ... 
arXiv:1905.01039v1 fatcat:rszzw7vb6nf2rc2z6nq43najdm

Perosonalized Differentially Private Location Collection Method with Adaptive GPS Discretization [chapter]

Huichuan Liu, Yong Zeng, Jiale Liu, Zhihong Liu, Jianfeng Ma, Xiaoyan Zhu
2020 Communications in Computer and Information Science  
To protect user privacy, researchers have adopted local differential privacy in data collection process.  ...  Thus in this paper, we design a differentially private location division module that could automatically discretize locations according to access density of each region.  ...  Local differential privacy will disturb each user's data, and the variance of the noise of the aggregate result is proportional to the number of samples.  ... 
doi:10.1007/978-981-33-4922-3_13 fatcat:d4velgugonboddipdvnitn3jtm


Sanjay Krishnan, Jiannan Wang, Michael J. Franklin, Ken Goldberg, Tim Kraska
2016 Proceedings of the 2016 International Conference on Management of Data - SIGMOD '16  
PrivateClean includes a technique for creating private datasets of numerical and discrete-valued attributes, a formalism for privacy-preserving data cleaning, and techniques for answering sum, count, and  ...  This paper explores the link between data cleaning and differential privacy in a framework we call PrivateClean.  ...  Aggregates over Select-Project-Join Views: PrivateClean can also be extended to estimate results for aggregate queries over SPJ views of differentially private relations.  ... 
doi:10.1145/2882903.2915248 dblp:conf/sigmod/KrishnanWFGK16 fatcat:rrfg5zrcofd4nphvflu5pmbpxu

D2P-Fed: Differentially Private Federated Learning With Efficient Communication [article]

Lun Wang, Ruoxi Jia, Dawn Song
2021 arXiv   pre-print
In this paper, we propose the discrete Gaussian based differentially private federated learning (D2P-Fed), a unified scheme to achieve both differential privacy (DP) and communication efficiency in federated  ...  The key idea is to apply the discrete Gaussian noise to the private data transmission. We provide complete analysis of the privacy guarantee, communication cost and convergence rate of D2P-Fed.  ...  In federated learning, each client performs training locally on their data source and only updates the model change to the server, which then updates the global model based on the aggregated local updates  ... 
arXiv:2006.13039v5 fatcat:sd4lbazazbgahdvidxwr5eoq7a

Towards Sparse Federated Analytics: Location Heatmaps under Distributed Differential Privacy with Secure Aggregation [article]

Eugene Bagdasaryan, Peter Kairouz, Stefan Mellem, Adrià Gascón, Kallista Bonawitz, Deborah Estrin, Marco Gruteser
2022 arXiv   pre-print
To achieve this, we revisit distributed differential privacy based on recent results in secure multiparty computation, and we design a scalable and adaptive distributed differential privacy approach for  ...  We design a scalable algorithm to privately generate location heatmaps over decentralized data from millions of user devices.  ...  Under distributed DP, clients first add noise to their data locally and then submit their noisy data to a private aggregation protocol.  ... 
arXiv:2111.02356v2 fatcat:5sgu3qkmfbejrknm2xtlpcb7wq

A Comprehensive Survey on Local Differential Privacy

Xingxing Xiong, Shubo Liu, Dan Li, Zhaohui Cai, Xiaoguang Niu
2020 Security and Communication Networks  
Local differential privacy (LDP) is a state-of-the-art privacy preservation technique that allows to perform big data analysis (e.g., statistical estimation, statistical learning, and data mining) while  ...  ., frequency estimation and mean estimation) and complex statistical estimation (e.g., multivariate distribution estimation and private estimation over complex data) under LDP.  ...  Some studies focused on the investigation of locally differentially private distribution estimation on the original data.  ... 
doi:10.1155/2020/8829523 fatcat:xjk3vgyambb5xioc2q5hyr2hua

Building a RAPPOR with the Unknown: Privacy-Preserving Learning of Associations and Data Dictionaries [article]

Giulia Fanti, Vasyl Pihur, Úlfar Erlingsson
2015 arXiv   pre-print
Techniques based on randomized response enable the collection of potentially sensitive data from clients in a privacy-preserving manner with strong local differential privacy guarantees.  ...  To enable learning without explicit knowledge of the dictionary, we develop methodology for estimating the joint distribution of two or more variables collected with RAPPOR.  ...  Our methods can be easily extended to any locally differentially-private system that is attempting to learn a distribution of discrete, string-valued random variables. II.  ... 
arXiv:1503.01214v1 fatcat:lslzwg3eerbovebufqjvdsjw4m

Distributed Differentially Private Algorithms for Matrix and Tensor Factorization

Hafiz Imtiaz, Anand D. Sarwate
2018 IEEE Journal on Selected Topics in Signal Processing  
This paper designs new and improved distributed and differentially private algorithms for two popular matrix and tensor factorization methods: principal component analysis (PCA) and orthogonal tensor decomposition  ...  In the distributed setting, differentially private algorithms suffer because they introduce noise to guarantee privacy.  ...  differentially-private OTD algorithm that does not employ correlated noise (conv), a differentially-private OTD [21] on local data (local) and the nonprivate tensor power method [3] on pooled data  ... 
doi:10.1109/jstsp.2018.2877842 pmid:31595179 pmcid:PMC6782067 fatcat:7tpb5h3p7nbazkw7remfvcwpzu

Building a RAPPOR with the Unknown: Privacy-Preserving Learning of Associations and Data Dictionaries

Giulia Fanti, Vasyl Pihur, Úlfar Erlingsson
2016 Proceedings on Privacy Enhancing Technologies  
Our contributions are not RAPPOR-specific, and can be generalized to other local differential privacy mechanisms for learning distributions of string-valued random variables.  ...  Techniques based on randomized response enable the collection of potentially sensitive data from clients in a privacy-preserving manner with strong local differential privacy guarantees.  ...  Special thanks to Andy Chu and Sooel Son for providing the Play Store data.  ... 
doi:10.1515/popets-2016-0015 dblp:journals/popets/FantiPE16 fatcat:l26qlewxbnggfgpsagtdwtzjp4

The Poisson Binomial Mechanism for Unbiased Federated Learning with Secure Aggregation

Wei-Ning Chen, Ayfer Özgür, Peter Kairouz
2022 International Conference on Machine Learning  
We introduce the Poisson Binomial mechanism (PBM), a discrete differential privacy mechanism for distributed mean estimation (DME) with applications to federated learning and analytics.  ...  estimator of the sum of the local vectors.  ...  The authors would like to thank Thomas Steinke for inspiring discussion in the early stage of this work.  ... 
dblp:conf/icml/ChenOK22 fatcat:wifr5wqsnndv7gevyufi6e5re4

Multi-Central Differential Privacy [article]

Thomas Steinke
2020 arXiv   pre-print
Differential privacy is typically studied in the central model where a trusted "aggregator" holds the sensitive data of all the individuals and is responsible for protecting their privacy.  ...  A popular alternative is the local model in which the aggregator is untrusted and instead each individual is responsible for their own privacy.  ...  For example, the European Union's General Data Protection Regulation (EU GDPR) restricts the transfer of private data abroad.  ... 
arXiv:2009.05401v1 fatcat:texputf4rvdablodxcy7byyfum

Local Differential Privacy and Its Applications: A Comprehensive Survey [article]

Mengmeng Yang, Lingjuan Lyu, Jun Zhao, Tianqing Zhu, Kwok-Yan Lam
2020 arXiv   pre-print
With the fast development of Information Technology, a tremendous amount of data have been generated and collected for research and analysis purposes.  ...  This survey provides a comprehensive and structured overview of the local differential privacy technology.  ...  Data aggregation under local differential privacy mainly includs four processes: Encoding, Perturbation, Aggregation, and Estimation.  ... 
arXiv:2008.03686v1 fatcat:l7z3gip2ivdmvin7lraxd4vciy

Improving Utility of Differentially Private Mechanisms through Cryptography-based Technologies: a Survey [article]

Wen Huang, Shijie Zhou, Tianqing Zhu, Yongjian Liao
2021 arXiv   pre-print
Next, we summarize hardness results of what is impossible to achieve for differentially private mechanisms' utility from the view of cryptography.  ...  Then, we summarize how to improve utility by combining differentially private mechanisms with homomorphic encryption schemes.  ...  In specific, each participant of local differential privacy mechanisms perturbs his data locally before these data are sent to an aggregator such that aggregation result satisfies requirements of differential  ... 
arXiv:2011.00976v2 fatcat:fizzcprz55cdxa7bwzyt7rzree
« Previous Showing results 1 — 15 out of 37,097 results