Please use this identifier to cite or link to this item: http://hdl.handle.net/10397/88239
Title: BDPL : a boundary differentially private layer against machine learning model extraction attacks
Authors: Zheng, H 
Ye, Q 
Hu, H 
Fang, C
Shi, J
Issue Date: 2019
Source: Lecture notes in computer science (including subseries Lecture notes in artificial intelligence and lecture notes in bioinformatics), 2019, v. 11735, p. 66-83
Abstract: Machine learning models trained by large volume of proprietary data and intensive computational resources are valuable assets of their owners, who merchandise these models to third-party users through prediction service API. However, existing literature shows that model parameters are vulnerable to extraction attacks which accumulate a large number of prediction queries and their responses to train a replica model. As countermeasures, researchers have proposed to reduce the rich API output, such as hiding the precise confidence level of the prediction response. Nonetheless, even with response being only one bit, an adversary can still exploit fine-tuned queries with differential property to infer the decision boundary of the underlying model. In this paper, we propose boundary differential privacy ( ϵ -BDP) as a solution to protect against such attacks by obfuscating the prediction responses near the decision boundary. ϵ -BDP guarantees an adversary cannot learn the decision boundary by a predefined precision no matter how many queries are issued to the prediction API. We design and prove a perturbation algorithm called boundary randomized response that can achieve ϵ -BDP. The effectiveness and high utility of our solution against model extraction attacks are verified by extensive experiments on both linear and non-linear models.
Publisher: Springer
Journal: Lecture notes in computer science (including subseries Lecture notes in artificial intelligence and lecture notes in bioinformatics) 
ISBN: 978-3-030-29958-3 (print)
978-3-030-29959-0 (online)
ISSN: 0302-9743
EISSN: 1611-3349
DOI: 10.1007/978-3-030-29959-0_4
Description: 24th European Symposium on Research in Computer Security, Luxembourg, September 23-27, 2019
Rights: © Springer Nature Switzerland AG 2019
Zheng H., Ye Q., Hu H., Fang C., Shi J. (2019) BDPL: A Boundary Differentially Private Layer Against Machine Learning Model Extraction Attacks. In: Sako K., Schneider S., Ryan P. (eds) Computer Security – ESORICS 2019. ESORICS 2019. Lecture Notes in Computer Science, vol 11735. Springer, Cham.
The final authenticated version is available online at https://doi.org/10.1007/978-3-030-29959-0_4.
Appears in Collections:Conference Paper

Files in This Item:
File Description SizeFormat 
ESORICS19.pdfPre-Published version1.23 MBAdobe PDFView/Open
Access
View full-text via PolyU eLinks SFX Query
Show full item record
PIRA download icon_1.1View/Download Full Text

Page view(s)

82
Citations as of Oct 15, 2020

Download(s)

10
Citations as of Oct 15, 2020

Google ScholarTM

Check

Altmetric


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.