research-article

Public Access

Dynamic Privacy Budget Allocation Improves Data Efficiency of Differentially Private Gradient Descent

Authors:

Zhangyang Wang,

Jiayu ZhouAuthors Info & Claims

FAccT '22: Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency

Pages 11 - 35

https://doi.org/10.1145/3531146.3533070

Published: 20 June 2022 Publication History

All formats PDF

Abstract

Protecting privacy in learning while maintaining the model performance has become increasingly critical in many applications that involve sensitive data. A popular private learning framework is differentially private learning composed of many privatized gradient iterations by noising and clipping. Under the privacy constraint, it has been shown that the dynamic policies could improve the final iterate loss, namely the quality of published models. In this talk, we will introduce these dynamic techniques for learning rate, batch size, noise magnitude and gradient clipping. Also, we discuss how the dynamic policy could change the convergence bounds which further provides insight of the impact of dynamic methods.

References

[1]

Martin Abadi, Andy Chu, Ian Goodfellow, H. Brendan McMahan, Ilya Mironov, Kunal Talwar, and Li Zhang. 2016. Deep Learning with Differential Privacy. In CCS: Proceedings of the 2016 ACM SIGSAC Conference on Computer and Communications Security(CCS ’16). ACM, New York, NY, USA, 308–318.

[2]

Raef Bassily, Vitaly Feldman, Kunal Talwar, and Abhradeep Guha Thakurta. 2019. Private Stochastic Convex Optimization with Optimal Rates. In Advances in Neural Information Processing Systems 32. Curran Associates, Inc., 11282–11291.

[3]

R. Bassily, A. Smith, and A. Thakurta. 2014. Private Empirical Risk Minimization: Efficient Algorithms and Tight Error Bounds. In 2014 IEEE 55th Annual Symposium on Foundations of Computer Science. 464–473.

[4]

Mark Bun and Thomas Steinke. 2016. Concentrated Differential Privacy: Simplifications, Extensions, and Lower Bounds. In Theory of Cryptography. Vol. 9985. Springer Berlin Heidelberg, Berlin, Heidelberg, 635–658.

[5]

Kamalika Chaudhuri, Claire Monteleoni, and Anand D. Sarwate. 2011. Differentially Private Empirical Risk Minimization. Journal of Machine Learning Research 12, Mar (2011), 1069–1109.

Digital Library

[6]

Junhong Cheng, Wenyan Liu, Xiaoling Wang, Xingjian Lu, Jing Feng, Yi Li, and Chaofan Duan. 2020. Adaptive Distributed Differential Privacy with SGD. Workshop on Privacy-Preserving Artificial Intelligence (2020), 6.

[7]

Rachel Cummings, Sara Krehbiel, Kevin A Lai, and Uthaipon Tantipongpipat. 2018. Differential Privacy for Growing Databases. In Advances in Neural Information Processing Systems 31. Curran Associates, Inc., 8864–8873.

[8]

Damien Desfontaines and Balázs Pejó. 2019. SoK: Differential Privacies. arXiv:1906.01337 [cs] (June 2019).

[9]

Cynthia Dwork, Alan Karr, Kobbi Nissim, and Lars Vilhuber. 2020. On Privacy in the Age of COVID-19. Journal of Privacy and Confidentiality 10, 2 (June 2020).

[10]

Cynthia Dwork, Frank McSherry, Kobbi Nissim, and Adam Smith. 2006. Calibrating Noise to Sensitivity in Private Data Analysis. In Theory of Cryptography(Lecture Notes in Computer Science). Springer Berlin Heidelberg, 265–284.

[11]

Vitaly Feldman, Tomer Koren, and Kunal Talwar. 2020. Private stochastic convex optimization: optimal rates in linear time. In Proceedings of the 52nd Annual ACM SIGACT Symposium on Theory of Computing(STOC 2020). Association for Computing Machinery, New York, NY, USA, 439–449.

Digital Library

[12]

Matt Fredrikson, Somesh Jha, and Thomas Ristenpart. 2015. Model Inversion Attacks That Exploit Confidence Information and Basic Countermeasures. In CCS: Proceedings of the 22Nd ACM SIGSAC Conference on Computer and Communications Security(CCS ’15). ACM, New York, NY, USA, 1322–1333.

[13]

Junyuan Hong, Haotao Wang, Zhangyang Wang, and Jiayu Zhou. 2021. Learning Model-Based Privacy Protection under Budget Constraints. In AAAI. 9.

[14]

Xixi Huang, Jian Guan, Bin Zhang, Shuhan Qi, Xuan Wang, and Qing Liao. 2019. Differentially Private Convolutional Neural Networks with Adaptive Gradient Descent. In 2019 IEEE Fourth International Conference on Data Science in Cyberspace (DSC). 642–648.

[15]

Prateek Jain, Dheeraj Nagaraj, and Praneeth Netrapalli. 2019. Making the Last Iterate of SGD Information Theoretically Optimal. In Conference on Learning Theory. 1752–1755.

[16]

Hamed Karimi, Julie Nutini, and Mark Schmidt. 2016. Linear Convergence of Gradient and Proximal-Gradient Methods Under the Polyak-Łojasiewicz Condition. In Machine Learning and Knowledge Discovery in Databases(Lecture Notes in Computer Science). Springer International Publishing, Cham, 795–811.

[17]

Diederik P. Kingma and Jimmy Ba. 2015. Adam: A Method for Stochastic Optimization. In the 3rd International Conference for Learning Representations. San Diego, CA.

[18]

Y. Lecun, L. Bottou, Y. Bengio, and P. Haffner. 1998. Gradient-based learning applied to document recognition. Proc. IEEE 86, 11 (Nov. 1998), 2278–2324.

[19]

Jaewoo Lee and Daniel Kifer. 2018. Concentrated Differentially Private Gradient Descent with Adaptive per-Iteration Privacy Budget. In Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining(KDD ’18). ACM, New York, NY, USA, 1656–1665.

Digital Library

[20]

Brendan McMahan, Eider Moore, Daniel Ramage, Seth Hampson, and Blaise Aguera y Arcas. 2017. Communication-Efficient Learning of Deep Networks from Decentralized Data. In Artificial Intelligence and Statistics. 1273–1282.

[21]

H. Brendan McMahan, Daniel Ramage, Kunal Talwar, and Li Zhang. 2018. Learning Differentially Private Recurrent Language Models. In International Conference on Learning Representations.

[22]

Yurii Nesterov and B.T. Polyak. 2006. Cubic regularization of Newton method and its global performance. Mathematical Programming 108, 1 (Aug. 2006), 177–205.

[23]

Venkatadheeraj Pichapati, Ananda Theertha Suresh, Felix X. Yu, Sashank J. Reddi, and Sanjiv Kumar. 2019. AdaCliP: Adaptive Clipping for Private SGD. arXiv:1908.07643 [cs, stat] (Oct. 2019).

[24]

B. T. Polyak. 1963. Gradient methods for the minimisation of functionals. U. S. S. R. Comput. Math. and Math. Phys. 3, 4 (Jan. 1963), 864–878.

[25]

B. T. Polyak. 1964. Some methods of speeding up the convergence of iteration methods. U. S. S. R. Comput. Math. and Math. Phys. 4, 5 (Jan. 1964), 1–17.

[26]

Ning Qian. 1999. On the momentum term in gradient descent learning algorithms. Neural Networks 12, 1 (Jan. 1999), 145–151.

Digital Library

[27]

Sashank J. Reddi, Ahmed Hefny, Suvrit Sra, Barnabas Poczos, and Alex Smola. 2016. Stochastic Variance Reduction for Nonconvex Optimization. In International Conference on Machine Learning. 314–323.

[28]

Alfréd Rényi. 1961. On measures of entropy and information. In Proceedings of the Fourth Berkeley Symposium on Mathematical Statistics and Probability, Volume 1: Contributions to the Theory of Statistics. The Regents of the University of California.

[29]

Shai Shalev-Shwartz and Shai Ben-David. 2014. Understanding Machine Learning: From Theory to Algorithms. Cambridge University Press.

Digital Library

[30]

Shai Shalev-Shwartz, Nathan Srebro, and Karthik Sridharan. 2009. Stochastic Convex Optimization. In Proceedings of the 22nd Annual Conference on Learning Theory, COLT ’09. 11.

[31]

R. Shokri, M. Stronati, C. Song, and V. Shmatikov. 2017. Membership Inference Attacks Against Machine Learning Models. In 2017 IEEE Symposium on Security and Privacy (SP). 3–18.

[32]

Om Thakkar, Galen Andrew, and H. Brendan McMahan. 2019. Differentially Private Learning with Adaptive Clipping. arXiv:1905.03871 [cs, stat] (May 2019).

[33]

Di Wang, Changyou Chen, and Jinhui Xu. 2019. Differentially Private Empirical Risk Minimization with Non-convex Loss Functions. In International Conference on Machine Learning. 6526–6535.

[34]

Di Wang, Minwei Ye, and Jinhui Xu. 2017. Differentially Private Empirical Risk Minimization Revisited: Faster and More General. In Advances in Neural Information Processing Systems 30. Curran Associates, Inc., 2722–2731.

[35]

Michael W. Weiner, Dallas P. Veitch, Paul S. Aisen, Laurel A. Beckett, Nigel J. Cairns, Robert C. Green, Danielle Harvey, Clifford R. Jack, William Jagust, Enchi Liu, John C. Morris, Ronald C. Petersen, Andrew J. Saykin, Mark E. Schmidt, Leslie Shaw, Li Shen, Judith A. Siuciak, Holly Soares, Arthur W. Toga, and John Q. Trojanowski. 2013. The Alzheimer’s Disease Neuroimaging Initiative: A review of papers published since its inception. Alzheimer’s & Dementia 9, 5 (Sept. 2013), e111–e194.

[36]

Xi Wu, Fengan Li, Arun Kumar, Kamalika Chaudhuri, Somesh Jha, and Jeffrey Naughton. 2017. Bolt-on Differential Privacy for Scalable Stochastic Gradient Descent-based Analytics. In Proceedings of the 2017 ACM International Conference on Management of Data(SIGMOD ’17). ACM, New York, NY, USA, 1307–1322.

Digital Library

[37]

Yun Xie, Peng Li, Chao Wu, and Qiuling Wu. 2021. Differential Privacy Stochastic Gradient Descent with Adaptive Privacy Budget Allocation. In 2021 IEEE International Conference on Consumer Electronics and Computer Engineering (ICCECE). 227–231.

[38]

Zhiying Xu, Shuyu Shi, Alex X. Liu, Jun Zhao, and Lin Chen. 2020. An Adaptive and Fast Convergent Approach to Differentially Private Deep Learning. the Proceedings of IEEE International Conference on Computer Communications (2020).

[39]

Da Yu, Huishuai Zhang, Wei Chen, Jian Yin, and Tie-Yan Liu. 2020. Gradient Perturbation is Underrated for Differentially Private Convex Optimization. In Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence. International Joint Conferences on Artificial Intelligence Organization, Yokohama, Japan, 3117–3123.

[40]

Lei Yu, Ling Liu, Calton Pu, Mehmet Emre Gursoy, and Stacey Truex. 2019. Differentially Private Model Publishing for Deep Learning. proceedings of 40th IEEE Symposium on Security and Privacy (April 2019).

[41]

Xinyue Zhang, Jiahao Ding, Maoqiang Wu, Stephen T. C. Wong, Hien Van Nguyen, and Miao Pan. 2021. Adaptive Privacy Preserving Deep Learning Algorithms for Medical Data. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. 1169–1178.

[42]

Yingxue Zhou, Xiangyi Chen, Mingyi Hong, Zhiwei Steven Wu, and Arindam Banerjee. 2020. Private Stochastic Non-Convex Optimization: Adaptive Algorithms and Tighter Generalization Bounds. arXiv:2006.13501 [cs, stat] (Aug. 2020).

Cited By

Hoang BPang YLiang SZhan LThompson PZhou JBaeza-Yates RBonchi F(2024)Distributed Harmonization: Federated Clustered Batch Effect Adjustment and GeneralizationProceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining10.1145/3637528.3671590(5105-5115)Online publication date: 25-Aug-2024
https://dl.acm.org/doi/10.1145/3637528.3671590

Index Terms

Dynamic Privacy Budget Allocation Improves Data Efficiency of Differentially Private Gradient Descent

Index terms have been assigned to the content through auto-classification.

Recommendations

Concentrated Differentially Private Gradient Descent with Adaptive per-Iteration Privacy Budget
KDD '18: Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining

Iterative algorithms, like gradient descent, are common tools for solving a variety of problems, such as model fitting. For this reason, there is interest in creating differentially private versions of them. However, their conversion to differentially ...
Differentially private regression analysis with dynamic privacy allocation
Abstract
In recent years, machine learning has reaped huge fruits in the domain of artificial intelligence. However, during the process of model training, machine learning models may be afflicted with the risk of disclosing sensitive ...
A Survey on Privacy Preserving Dynamic Data Publishing

Many organizations, especially small and medium business SMB enterprises require the collection and sharing of data containing personal information. The privacy of this data must be preserved before outsourcing to the commercial public. Privacy ...

Comments

Information & Contributors

Information

Published In

cover image ACM Other conferences

FAccT '22: Proceedings of the 2022 ACM Conference on Fairness, Accountability, and Transparency

June 2022

2351 pages

ISBN:9781450393522

DOI:10.1145/3531146

Copyright © 2022 ACM.

Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from [email protected]

Sponsors

ACM: Association for Computing Machinery

Publisher

Association for Computing Machinery

New York, NY, United States

Publication History

Published: 20 June 2022

Permissions

Request permissions for this article.

Request Permissions

Check for updates

Author Tags

Qualifiers

Research-article
Research
Refereed limited

Funding Sources

National Institute of Aging
Office of Naval Research
Good Systems
National Science Foundation

Conference

FAccT '22

Sponsor:

ACM

FAccT '22: 2022 ACM Conference on Fairness, Accountability, and Transparency

June 21 - 24, 2022

Seoul, Republic of Korea

Contributors

Other Metrics

View Article Metrics

Bibliometrics & Citations

Bibliometrics

Article Metrics

1
Total Citations
View Citations
604
Total Downloads

Downloads (Last 12 months)375
Downloads (Last 6 weeks)82

Reflects downloads up to 30 Oct 2024

Other Metrics

View Author Metrics

Citations

Cited By

Hoang BPang YLiang SZhan LThompson PZhou JBaeza-Yates RBonchi F(2024)Distributed Harmonization: Federated Clustered Batch Effect Adjustment and GeneralizationProceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining10.1145/3637528.3671590(5105-5115)Online publication date: 25-Aug-2024
https://dl.acm.org/doi/10.1145/3637528.3671590

View Options

View options

PDF

View or Download as a PDF file.

eReader

View online with eReader.

HTML Format

View this article in HTML Format.

Get Access

Login options

Check if you have access through your login credentials or your institution to get full access on this article.

Full Access

Get this Publication

Media

Figures

Other

Tables

View Table of Contents