Privacy risks of securing machine learning models against adversarial examples

Liwei Song, Reza Shokri, Prateek Mittal

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Scopus citations

Abstract

The arms race between attacks and defenses for machine learning models has come to a forefront in recent years, in both the security community and the privacy community. However, one big limitation of previous research is that the security domain and the privacy domain have typically been considered separately. It is thus unclear whether the defense methods in one domain will have any unexpected impact on the other domain. In this paper, we take a step towards resolving this limitation by combining the two domains. In particular, we measure the success of membership inference attacks against six state-of-the-art defense methods that mitigate the risk of adversarial examples (i.e., evasion attacks). Membership inference attacks determine whether or not an individual data record has been part of a model's training set. The accuracy of such attacks reflects the information leakage of training algorithms about individual members of the training set. Adversarial defense methods against adversarial examples influence the model's decision boundaries such that model predictions remain unchanged for a small area around each input. However, this objective is optimized on training data. Thus, individual data records in the training set have a significant influence on robust models. This makes the models more vulnerable to inference attacks. To perform the membership inference attacks, we leverage the existing inference methods that exploit model predictions. We also propose two new inference methods that exploit structural properties of robust models on adversarially perturbed data. Our experimental evaluation demonstrates that compared with the natural training (undefended) approach, adversarial defense methods can indeed increase the target model's risk against membership inference attacks. When using adversarial defenses to train the robust models, the membership inference advantage increases by up to 4.5 times compared to the naturally undefended models. Beyond revealing the privacy risks of adversarial defenses, we further investigate the factors, such as model capacity, that influence the membership information leakage.

Original languageEnglish (US)
Title of host publicationCCS 2019 - Proceedings of the 2019 ACM SIGSAC Conference on Computer and Communications Security
PublisherAssociation for Computing Machinery
Pages241-257
Number of pages17
ISBN (Electronic)9781450367479
DOIs
StatePublished - Nov 6 2019
Event26th ACM SIGSAC Conference on Computer and Communications Security, CCS 2019 - London, United Kingdom
Duration: Nov 11 2019Nov 15 2019

Publication series

NameProceedings of the ACM Conference on Computer and Communications Security
ISSN (Print)1543-7221

Conference

Conference26th ACM SIGSAC Conference on Computer and Communications Security, CCS 2019
CountryUnited Kingdom
CityLondon
Period11/11/1911/15/19

All Science Journal Classification (ASJC) codes

  • Software
  • Computer Networks and Communications

Keywords

  • Adversarial examples
  • Defenses
  • Machine learning
  • Membership inference attacks

Fingerprint Dive into the research topics of 'Privacy risks of securing machine learning models against adversarial examples'. Together they form a unique fingerprint.

  • Cite this

    Song, L., Shokri, R., & Mittal, P. (2019). Privacy risks of securing machine learning models against adversarial examples. In CCS 2019 - Proceedings of the 2019 ACM SIGSAC Conference on Computer and Communications Security (pp. 241-257). (Proceedings of the ACM Conference on Computer and Communications Security). Association for Computing Machinery. https://doi.org/10.1145/3319535.3354211