A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2019; you can also visit the original URL.
The file type is application/pdf.
Filters
Differentially Private Data Generative Models
[article]
2018
arXiv
pre-print
Preserved Fulltext
differentially private variational autoencoder-based generative model (DP-VaeGM). ...
In this paper, to enable learning efficiency as well as to generate data with privacy guarantees and high utility, we propose a differentially private autoencoder-based generative model (DP-AuGM) and a ...
Let M denote the differentially private generative model and X be the private data. ...
arXiv:1812.02274v1
fatcat:apvq4zrl7rfuvmnlkphaowe4f4
A Latent Class Modeling Approach for Generating Synthetic Data and Making Posterior Inferences from Differentially Private Counts
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
We present a latent class modeling approach for post-processing differentially private marginal counts that can be used (i) to create differentially private synthetic data from the set of marginal counts ...
Several algorithms exist for creating differentially private counts from contingency tables, such as two-way or three-way marginal counts. ...
Acknowledgment This work was supported by the Pennsylvania State University, Duke University, ACI under award 1443014, and the National Science Foundation under an IGERT award #DGE-1144860, Big Data Social ...
arXiv:2201.10545v1
fatcat:fbruzaf6p5c2jlbebsodasilhi
DP-CGAN: Differentially Private Synthetic Data and Label Generation
[article]
2020
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Generative Adversarial Networks (GANs) are one of the well-known models to generate synthetic data including images, especially for research communities that cannot use original sensitive datasets because ...
To address this challenge, we introduce a Differentially Private Conditional GAN (DP-CGAN) training framework based on a new clipping and perturbation strategy, which improves the performance of the model ...
Conclusion In this research, we proposed DP-CGAN framework that is a differentially private GAN model capable of generating both synthetic data and corresponding labels. ...
arXiv:2001.09700v1
fatcat:27ohelg7i5gpxffqvfnrfoenv4
Private Federated Learning with Domain Adaptation
[article]
2019
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
We show that this technique improves model accuracy for all users, using both real and synthetic data, and that this improvement is much more pronounced when differential privacy bounds are imposed on ...
Federated Learning (FL) is a distributed machine learning (ML) paradigm that enables multiple parties to jointly re-train a shared model without sharing their data with any other parties, offering advantages ...
For privacy in the general model, we use FL with differentially private stochastic gradient descent (SGD) (1). ...
arXiv:1912.06733v1
fatcat:csrii27ugbdjbau6xgzpvnihau
Differentially Private Synthetic Data: Applied Evaluations and Enhancements
[article]
2020
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Differentially private data synthesis protects personal details from exposure, and allows for the training of differentially private machine learning models on privately generated datasets. ...
But how can we effectively assess the efficacy of differentially private synthetic data? In this paper, we survey four differentially private generative adversarial networks for data synthesis. ...
the privatized data, and train informed supervised learning models. ...
arXiv:2011.05537v1
fatcat:ozosb6sjevhlnd673tk5vwmnuq
Aggregation for Privately Trained Different Types of Local Models
2020
EAI Endorsed Transactions on Security and Safety
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
To do so, firstly, we propose differentially private GANs, let local parties generate synthetic data related to their training data. ...
By combining synthetic data and labels from local models, knowledge can be transferred from local models to the global model. ...
By combining generating differentially private synthetic data and querying local models' predictions, we transfer knowledge from local models to the global model. ...
doi:10.4108/eai.21-6-2021.170237
fatcat:mqu25uq5preglpnyg2ktgg476u
Really Useful Synthetic Data – A Framework to Evaluate the Quality of Differentially Private Synthetic Data
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Other Versions
Acknowledging that data quality is a subjective concept, we develop a framework to evaluate the quality of differentially private synthetic data from an applied researcher's perspective. ...
Recent advances in generating synthetic data that allow to add principled ways of protecting privacy -- such as Differential Privacy -- are a crucial step in sharing statistical information in a privacy ...
the data that was used to generate the synthetic data) and the generalization properties of the differentially private synthetic data. ...
arXiv:2004.07740v2
fatcat:ar35f4jc75frnbwax3pwgzi32u
A Survey on Differentially Private Machine Learning [Review Article]
2020
IEEE Computational Intelligence Magazine
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Thus, analysts can use the welltrained private generative model to produce highquality data. ...
As a result, we can learn private generative models that provide differential privacy for each individual in the training data and generate realistic synthetic samples.
5) Summary In Table V of the ...
doi:10.1109/mci.2020.2976185
fatcat:72lxqnelszctfghmhdhdi3bk44
Not Just Cloud Privacy: Protecting Client Privacy in Teacher-Student Learning
[article]
2020
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2020; you can also visit the original URL.
The file type is application/pdf.
Note that this fulltext copy is not of the "primary" version of this work. The version it corresponds to is:
2019 pre-print arXiv:1910.08038v1 fatcat:i2pwnmmeiveiph7usvfgjxwvvy
In this work, we re-design the privacy-preserving "teacher-student" model consisting of adopting both private arbitrary masking and local differential privacy, which protects the sensitive information ...
However, the traditional training of teacher model is not robust on any perturbed data. ...
Adversarial Sample Generation: We generate the local differential private samples randomly from the teacher original data. ...
arXiv:1910.08038v2
fatcat:6xu7fcftjncgxleadbnkvhbvcu
Differentially Private Continual Learning
[article]
2019
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2019; you can also visit the original URL.
The file type is application/pdf.
We estimate the likelihood of past data given the current model using differentially private generative models of old datasets. ...
But neural networks trained on recent data alone will tend to forget lessons learned on old data. We present a differentially private continual learning framework based on variational inference. ...
Differentially Private Variational Generative Experience Replay (DP-VGER) is a dual-memory learning system that attempts to resolve this problem using differentially private generative models of old data ...
arXiv:1902.06497v1
fatcat:6h2qhhkk65fkflk7qmqolgvauy
Differentially Private Generative Adversarial Networks with Model Inversion
2021
2021 IEEE International Workshop on Information Forensics and Security (WIFS)
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Note that this fulltext copy is not of the "primary" version of this work. The version it corresponds to is:
2022 pre-print arXiv:2201.03139v1 fatcat:j7pfhp7265dn5dn6mc6vwxxhfm
We propose Differentially Private Model Inversion (DPMI) method where the private data is first mapped to the latent space via a public generator, followed by a lower-dimensional DP-GAN with better convergent ...
To protect sensitive data in training a Generative Adversarial Network (GAN), the standard approach is to use differentially private (DP) stochastic gradient descent method in which controlled noise is ...
CONCLUSIONS In this paper, we have introduced DPMI, a new differentially private generative framework of releasing synthetic private data by applying model inversion to map the real private data to the ...
doi:10.1109/wifs53200.2021.9648378
pmid:35517057
pmcid:PMC9070036
fatcat:o27xsdy3tzg3he5zv6awhciaam
DPD-InfoGAN: Differentially Private Distributed InfoGAN
[article]
2021
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2021; you can also visit the original URL.
The file type is application/pdf.
Other Versions
Due to the high model complexity of InfoGAN, the generative distribution tends to be concentrated around the training data points. ...
To address this problem, we propose a differentially private version of InfoGAN (DP-InfoGAN). ...
Similarly, the generator satisfies differential privacy, as the generator receives updates from the discriminator and the Q network which are trained in a differentially private manner. ...
arXiv:2010.11398v3
fatcat:2ossf23oprfdpgmtr6y6jgq3oy
Differential Privacy: What is all the noise about?
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
Differential Privacy (DP) is a formal definition of privacy that provides rigorous guarantees against risks of privacy breaches during data processing. ...
can be used to train differentially private models. ...
The algorithms privately learn to discriminate fake from real labels using both private and generated data from the public generator. ...
arXiv:2205.09453v1
fatcat:5z3nqsh7qbbwfhbrc6hmzt43ya
MC-GEN:Multi-level Clustering for Private Synthetic Data Generation
[article]
2022
arXiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf.
MC-GEN builds differentially private generative models on the multi-level clustered data to generate synthetic datasets. ...
We propose MC-GEN, a privacy-preserving synthetic data generation method under differential privacy guarantee for multiple classification tasks. ...
extracted statistical information. • Data generator: Generate synthetic data sample by sample from perturbed generative model. ...
arXiv:2205.14298v1
fatcat:u43pz5pocfe3rgt4zj2nmsoloa
Privacy-preserving generative deep neural networks support clinical data sharing
[article]
2017
bioRxiv
pre-print
Preserved Fulltext
A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2019; you can also visit the original URL.
The file type is application/pdf.
Using the SPRINT trial as an example, we show that machine-learning models built from simulated participants generalize to the original dataset. ...
Generated data can be released alongside analytical code to enable fully reproducible workflows, even when privacy is a concern. ...
Addition of differential 218 privacy during the synthetic data generation process (i.e., the "private dataset") generated data 219 generally reflecting these trends, but with an increased level of noise ...
doi:10.1101/159756
fatcat:beyeuvnwdfb2fai76we72kgnq4
« Previous
Showing results 1 — 15 out of 473,578 results