Optimal Balance of Privacy and Utility with Differential Privacy Deep Learning Frameworks

Olivera Kotevska; Folami Alamudun; Christopher Stanley

doi:10.1109/CSCI54926.2021.00141

Optimal Balance of Privacy and Utility with Differential Privacy Deep Learning Frameworks

Olivera Kotevska, Folami Alamudun, Christopher Stanley

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

2 Scopus citations

Abstract

As the number of online services has increased, the amount of sensitive data being recorded is rising. Simultaneously, the decision-making process has improved by using the vast amounts of data, where machine learning has transformed entire industries. This paper addresses the development of optimal private deep neural networks and discusses the challenges associated with this task. We focus on differential privacy implementations and finding the optimal balance between accuracy and privacy, benefits and limitations of existing libraries, and challenges of applying private machine learning models in practical applications. Our analysis shows that learning rate, and privacy budget are the key factors that impact the results, and we discuss options for these settings.

Original language	English
Title of host publication	Proceedings - 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	425-430
Number of pages	6
ISBN (Electronic)	9781665458412
DOIs	https://doi.org/10.1109/CSCI54926.2021.00141
State	Published - 2021
Event	2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021 - Las Vegas, United States Duration: Dec 15 2021 → Dec 17 2021

Publication series

Name	Proceedings - 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021

Conference

Conference	2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021
Country/Territory	United States
City	Las Vegas
Period	12/15/21 → 12/17/21

Funding

This manuscript has been authored by UT-Battelle, LLC, under contract DE-AC05-00OR22725 with the US Department of Energy (DOE). The publisher acknowledges the US government license to provide public access under the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan). This research is sponsored by the Laboratory Directed Research and Development Program of Oak Ridge National Laboratory, managed by UT-Battelle,LLC, for the U.S. Department of Energy under contract DE-AC05-00OR22725. This material is also based upon work supported by the Department of Energy, Office of Science, Office of Advanced Scientific Computing Research.

Keywords

deep neural network
differential privacy
personal data
privacy

Access to Document

10.1109/CSCI54926.2021.00141

Cite this

Kotevska, O., Alamudun, F., & Stanley, C. (2021). Optimal Balance of Privacy and Utility with Differential Privacy Deep Learning Frameworks. In Proceedings - 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021 (pp. 425-430). (Proceedings - 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/CSCI54926.2021.00141

Kotevska, Olivera ; Alamudun, Folami ; Stanley, Christopher. / Optimal Balance of Privacy and Utility with Differential Privacy Deep Learning Frameworks. Proceedings - 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021. Institute of Electrical and Electronics Engineers Inc., 2021. pp. 425-430 (Proceedings - 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021).

@inproceedings{e3f63f73328a4dc3a29015bd9ffb97f2,

title = "Optimal Balance of Privacy and Utility with Differential Privacy Deep Learning Frameworks",

abstract = "As the number of online services has increased, the amount of sensitive data being recorded is rising. Simultaneously, the decision-making process has improved by using the vast amounts of data, where machine learning has transformed entire industries. This paper addresses the development of optimal private deep neural networks and discusses the challenges associated with this task. We focus on differential privacy implementations and finding the optimal balance between accuracy and privacy, benefits and limitations of existing libraries, and challenges of applying private machine learning models in practical applications. Our analysis shows that learning rate, and privacy budget are the key factors that impact the results, and we discuss options for these settings.",

keywords = "deep neural network, differential privacy, personal data, privacy",

author = "Olivera Kotevska and Folami Alamudun and Christopher Stanley",

note = "Publisher Copyright: {\textcopyright} 2021 IEEE.; 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021 ; Conference date: 15-12-2021 Through 17-12-2021",

year = "2021",

doi = "10.1109/CSCI54926.2021.00141",

language = "English",

series = "Proceedings - 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "425--430",

booktitle = "Proceedings - 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021",

}

Kotevska, O, Alamudun, F & Stanley, C 2021, Optimal Balance of Privacy and Utility with Differential Privacy Deep Learning Frameworks. in Proceedings - 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021. Proceedings - 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021, Institute of Electrical and Electronics Engineers Inc., pp. 425-430, 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021, Las Vegas, United States, 12/15/21. https://doi.org/10.1109/CSCI54926.2021.00141

Optimal Balance of Privacy and Utility with Differential Privacy Deep Learning Frameworks. / Kotevska, Olivera; Alamudun, Folami; Stanley, Christopher.
Proceedings - 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021. Institute of Electrical and Electronics Engineers Inc., 2021. p. 425-430 (Proceedings - 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Optimal Balance of Privacy and Utility with Differential Privacy Deep Learning Frameworks

AU - Kotevska, Olivera

AU - Alamudun, Folami

AU - Stanley, Christopher

PY - 2021

Y1 - 2021

N2 - As the number of online services has increased, the amount of sensitive data being recorded is rising. Simultaneously, the decision-making process has improved by using the vast amounts of data, where machine learning has transformed entire industries. This paper addresses the development of optimal private deep neural networks and discusses the challenges associated with this task. We focus on differential privacy implementations and finding the optimal balance between accuracy and privacy, benefits and limitations of existing libraries, and challenges of applying private machine learning models in practical applications. Our analysis shows that learning rate, and privacy budget are the key factors that impact the results, and we discuss options for these settings.

AB - As the number of online services has increased, the amount of sensitive data being recorded is rising. Simultaneously, the decision-making process has improved by using the vast amounts of data, where machine learning has transformed entire industries. This paper addresses the development of optimal private deep neural networks and discusses the challenges associated with this task. We focus on differential privacy implementations and finding the optimal balance between accuracy and privacy, benefits and limitations of existing libraries, and challenges of applying private machine learning models in practical applications. Our analysis shows that learning rate, and privacy budget are the key factors that impact the results, and we discuss options for these settings.

KW - deep neural network

KW - differential privacy

KW - personal data

KW - privacy

UR - http://www.scopus.com/inward/record.url?scp=85133915649&partnerID=8YFLogxK

U2 - 10.1109/CSCI54926.2021.00141

DO - 10.1109/CSCI54926.2021.00141

M3 - Conference contribution

AN - SCOPUS:85133915649

T3 - Proceedings - 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021

SP - 425

EP - 430

BT - Proceedings - 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021

Y2 - 15 December 2021 through 17 December 2021

ER -

Kotevska O, Alamudun F, Stanley C. Optimal Balance of Privacy and Utility with Differential Privacy Deep Learning Frameworks. In Proceedings - 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021. Institute of Electrical and Electronics Engineers Inc. 2021. p. 425-430. (Proceedings - 2021 International Conference on Computational Science and Computational Intelligence, CSCI 2021). doi: 10.1109/CSCI54926.2021.00141

Optimal Balance of Privacy and Utility with Differential Privacy Deep Learning Frameworks

Abstract

Publication series

Conference

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this