Privacy-preserving data generation and sharing using identification sanitizer

Shuo Wang; Lingjuan Lyu; Tianle Chen; Shangyu Chen; Surya Nepal; Carsten Rudolph; Marthie Grobler

doi:10.1007/978-3-030-62008-0_13

Privacy-preserving data generation and sharing using identification sanitizer

Shuo Wang
, Lingjuan Lyu
, Tianle Chen
, Shangyu Chen
, Surya Nepal
, Carsten Rudolph
, Marthie Grobler

Department of Software Systems & Cybersecurity

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

Abstract

In this paper, we propose a practical privacy-preserving generative model for data sanitization and sharing, called Sanitizer-Variational Autoencoder (SVAE). We assume that the data consists of identification-relevant and irrelevant components. A variational autoencoder (VAE) based sanitization model is proposed to strip the identification-relevant features and only retain identification-irrelevant components in a privacy-preserving manner. The sanitization allows for task-relevant discrimination (utility) but minimizes the personal identification information leakage (privacy). We conduct extensive empirical evaluations on the real-world face, biometric signal and speech datasets, and validate the effectiveness of our proposed SVAE, as well as the robustness against the membership inference attack.

Original language	English
Title of host publication	Web Information Systems Engineering – WISE 2020
Subtitle of host publication	21st International Conference Amsterdam, The Netherlands, October 20–24, 2020 Proceedings, Part II
Editors	Zhisheng Huang, Wouter Beek, Hua Wang, Rui Zhou, Yanchun Zhang
Place of Publication	Cham Switzerland
Publisher	Springer
Pages	185-200
Number of pages	16
ISBN (Electronic)	9783030620080
ISBN (Print)	9783030620073
DOIs	https://doi.org/10.1007/978-3-030-62008-0_13
Publication status	Published - 2020
Event	International Conference on Web Information Systems Engineering 2020 - Amsterdam, Netherlands Duration: 20 Oct 2020 → 24 Oct 2020 Conference number: 21st https://link.springer.com/book/10.1007/978-3-030-62008-0 (Proceedings) http://wasp.cs.vu.nl/WISE2020/ (Website)

Publication series

Name	Lecture Notes in Computer Science
Publisher	Springer
Number	Part II
Volume	12343
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	International Conference on Web Information Systems Engineering 2020
Abbreviated title	WISE 2020
Country/Territory	Netherlands
City	Amsterdam
Period	20/10/20 → 24/10/20
Internet address	https://link.springer.com/book/10.1007/978-3-030-62008-0 (Proceedings) http://wasp.cs.vu.nl/WISE2020/ (Website)

Keywords

Data sharing
Deep learning
Generative model
Privacy-preserving
Variational autoencoder

Access to Document

10.1007/978-3-030-62008-0_13

Cite this

Wang, S., Lyu, L., Chen, T., Chen, S., Nepal, S., Rudolph, C., & Grobler, M. (2020). Privacy-preserving data generation and sharing using identification sanitizer. In Z. Huang, W. Beek, H. Wang, R. Zhou, & Y. Zhang (Eds.), Web Information Systems Engineering – WISE 2020: 21st International Conference Amsterdam, The Netherlands, October 20–24, 2020 Proceedings, Part II (pp. 185-200). (Lecture Notes in Computer Science ; Vol. 12343 , No. Part II). Springer. https://doi.org/10.1007/978-3-030-62008-0_13

Wang, Shuo ; Lyu, Lingjuan ; Chen, Tianle et al. / Privacy-preserving data generation and sharing using identification sanitizer. Web Information Systems Engineering – WISE 2020: 21st International Conference Amsterdam, The Netherlands, October 20–24, 2020 Proceedings, Part II. editor / Zhisheng Huang ; Wouter Beek ; Hua Wang ; Rui Zhou ; Yanchun Zhang. Cham Switzerland : Springer, 2020. pp. 185-200 (Lecture Notes in Computer Science ; Part II).

@inproceedings{46dab61e542a4e24b8491f94c8c22409,

title = "Privacy-preserving data generation and sharing using identification sanitizer",

abstract = "In this paper, we propose a practical privacy-preserving generative model for data sanitization and sharing, called Sanitizer-Variational Autoencoder (SVAE). We assume that the data consists of identification-relevant and irrelevant components. A variational autoencoder (VAE) based sanitization model is proposed to strip the identification-relevant features and only retain identification-irrelevant components in a privacy-preserving manner. The sanitization allows for task-relevant discrimination (utility) but minimizes the personal identification information leakage (privacy). We conduct extensive empirical evaluations on the real-world face, biometric signal and speech datasets, and validate the effectiveness of our proposed SVAE, as well as the robustness against the membership inference attack.",

keywords = "Data sharing, Deep learning, Generative model, Privacy-preserving, Variational autoencoder",

author = "Shuo Wang and Lingjuan Lyu and Tianle Chen and Shangyu Chen and Surya Nepal and Carsten Rudolph and Marthie Grobler",

year = "2020",

doi = "10.1007/978-3-030-62008-0\_13",

language = "English",

isbn = "9783030620073",

series = "Lecture Notes in Computer Science ",

publisher = "Springer",

number = "Part II",

pages = "185--200",

editor = "Zhisheng Huang and Wouter Beek and Hua Wang and Rui Zhou and Yanchun Zhang",

booktitle = "Web Information Systems Engineering – WISE 2020",

address = "Switzerland",

note = "International Conference on Web Information Systems Engineering 2020, WISE 2020 ; Conference date: 20-10-2020 Through 24-10-2020",

url = "https://link.springer.com/book/10.1007/978-3-030-62008-0, http://wasp.cs.vu.nl/WISE2020/",

}

Wang, S, Lyu, L, Chen, T, Chen, S, Nepal, S, Rudolph, C & Grobler, M 2020, Privacy-preserving data generation and sharing using identification sanitizer. in Z Huang, W Beek, H Wang, R Zhou & Y Zhang (eds), Web Information Systems Engineering – WISE 2020: 21st International Conference Amsterdam, The Netherlands, October 20–24, 2020 Proceedings, Part II. Lecture Notes in Computer Science , no. Part II, vol. 12343 , Springer, Cham Switzerland, pp. 185-200, International Conference on Web Information Systems Engineering 2020, Amsterdam, Netherlands, 20/10/20. https://doi.org/10.1007/978-3-030-62008-0_13

Privacy-preserving data generation and sharing using identification sanitizer. / Wang, Shuo; Lyu, Lingjuan; Chen, Tianle et al.
Web Information Systems Engineering – WISE 2020: 21st International Conference Amsterdam, The Netherlands, October 20–24, 2020 Proceedings, Part II. ed. / Zhisheng Huang; Wouter Beek; Hua Wang; Rui Zhou; Yanchun Zhang. Cham Switzerland: Springer, 2020. p. 185-200 (Lecture Notes in Computer Science ; Vol. 12343 , No. Part II).

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

TY - GEN

T1 - Privacy-preserving data generation and sharing using identification sanitizer

AU - Wang, Shuo

AU - Lyu, Lingjuan

AU - Chen, Tianle

AU - Chen, Shangyu

AU - Nepal, Surya

AU - Rudolph, Carsten

AU - Grobler, Marthie

N1 - Conference code: 21st

PY - 2020

Y1 - 2020

N2 - In this paper, we propose a practical privacy-preserving generative model for data sanitization and sharing, called Sanitizer-Variational Autoencoder (SVAE). We assume that the data consists of identification-relevant and irrelevant components. A variational autoencoder (VAE) based sanitization model is proposed to strip the identification-relevant features and only retain identification-irrelevant components in a privacy-preserving manner. The sanitization allows for task-relevant discrimination (utility) but minimizes the personal identification information leakage (privacy). We conduct extensive empirical evaluations on the real-world face, biometric signal and speech datasets, and validate the effectiveness of our proposed SVAE, as well as the robustness against the membership inference attack.

AB - In this paper, we propose a practical privacy-preserving generative model for data sanitization and sharing, called Sanitizer-Variational Autoencoder (SVAE). We assume that the data consists of identification-relevant and irrelevant components. A variational autoencoder (VAE) based sanitization model is proposed to strip the identification-relevant features and only retain identification-irrelevant components in a privacy-preserving manner. The sanitization allows for task-relevant discrimination (utility) but minimizes the personal identification information leakage (privacy). We conduct extensive empirical evaluations on the real-world face, biometric signal and speech datasets, and validate the effectiveness of our proposed SVAE, as well as the robustness against the membership inference attack.

KW - Data sharing

KW - Deep learning

KW - Generative model

KW - Privacy-preserving

KW - Variational autoencoder

UR - https://www.scopus.com/pages/publications/85096620522

U2 - 10.1007/978-3-030-62008-0_13

DO - 10.1007/978-3-030-62008-0_13

M3 - Conference Paper

AN - SCOPUS:85096620522

SN - 9783030620073

T3 - Lecture Notes in Computer Science

SP - 185

EP - 200

BT - Web Information Systems Engineering – WISE 2020

A2 - Huang, Zhisheng

A2 - Beek, Wouter

A2 - Wang, Hua

A2 - Zhou, Rui

A2 - Zhang, Yanchun

PB - Springer

CY - Cham Switzerland

T2 - International Conference on Web Information Systems Engineering 2020

Y2 - 20 October 2020 through 24 October 2020

ER -

Wang S, Lyu L, Chen T, Chen S, Nepal S, Rudolph C et al. Privacy-preserving data generation and sharing using identification sanitizer. In Huang Z, Beek W, Wang H, Zhou R, Zhang Y, editors, Web Information Systems Engineering – WISE 2020: 21st International Conference Amsterdam, The Netherlands, October 20–24, 2020 Proceedings, Part II. Cham Switzerland: Springer. 2020. p. 185-200. (Lecture Notes in Computer Science ; Part II). doi: 10.1007/978-3-030-62008-0_13

Privacy-preserving data generation and sharing using identification sanitizer

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Cite this