Current browse context:
cs.CR
Change to browse by:
References & Citations
Computer Science > Cryptography and Security
Title: Privacy Risk in Machine Learning: Analyzing the Connection to Overfitting
(Submitted on 5 Sep 2017 (v1), last revised 4 May 2018 (this version, v5))
Abstract: Machine learning algorithms, when applied to sensitive data, pose a distinct threat to privacy. A growing body of prior work demonstrates that models produced by these algorithms may leak specific private information in the training data to an attacker, either through the models' structure or their observable behavior. However, the underlying cause of this privacy risk is not well understood beyond a handful of anecdotal accounts that suggest overfitting and influence might play a role.
This paper examines the effect that overfitting and influence have on the ability of an attacker to learn information about the training data from machine learning models, either through training set membership inference or attribute inference attacks. Using both formal and empirical analyses, we illustrate a clear relationship between these factors and the privacy risk that arises in several popular machine learning algorithms. We find that overfitting is sufficient to allow an attacker to perform membership inference and, when the target attribute meets certain conditions about its influence, attribute inference attacks. Interestingly, our formal analysis also shows that overfitting is not necessary for these attacks and begins to shed light on what other factors may be in play. Finally, we explore the connection between membership inference and attribute inference, showing that there are deep connections between the two that lead to effective new attacks.
Submission history
From: Samuel Yeom [view email][v1] Tue, 5 Sep 2017 21:56:24 GMT (731kb,D)
[v2] Thu, 7 Sep 2017 21:39:21 GMT (772kb,D)
[v3] Thu, 1 Feb 2018 03:38:13 GMT (458kb,D)
[v4] Fri, 2 Feb 2018 23:04:31 GMT (458kb,D)
[v5] Fri, 4 May 2018 22:43:43 GMT (457kb,D)
Link back to: arXiv, form interface, contact.