研究数据依赖下的成员推理攻击

Thomas Humphries, Simon Oya, Lindsey Tulloch, Matthew Rafuse, I. Goldberg, U. Hengartner, F. Kerschbaum
{"title":"研究数据依赖下的成员推理攻击","authors":"Thomas Humphries, Simon Oya, Lindsey Tulloch, Matthew Rafuse, I. Goldberg, U. Hengartner, F. Kerschbaum","doi":"10.1109/CSF57540.2023.00013","DOIUrl":null,"url":null,"abstract":"Training machine learning models on privacy-sensitive data has become a popular practice, driving innovation in ever-expanding fields. This has opened the door to new attacks that can have serious privacy implications. One such attack, the Membership Inference Attack (MIA), exposes whether or not a particular data point was used to train a model. A growing body of literature uses Differentially Private (DP) training algorithms as a defence against such attacks. However, these works evaluate the defence under the restrictive assumption that all members of the training set, as well as non-members, are independent and identically distributed. This assumption does not hold for many real-world use cases in the literature. Motivated by this, we evaluate membership inference with statistical dependencies among samples and explain why DP does not provide meaningful protection (the privacy parameter $\\epsilon$ scales with the training set size $n$) in this more general case. We conduct a series of empirical evaluations with off-the-shelf MIAs using training sets built from real-world data showing different types of dependencies among samples. Our results reveal that training set dependencies can severely increase the performance of MIAs, and therefore assuming that data samples are statistically independent can significantly underestimate the performance of MIAs.","PeriodicalId":179870,"journal":{"name":"2023 IEEE 36th Computer Security Foundations Symposium (CSF)","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2020-10-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"12","resultStr":"{\"title\":\"Investigating Membership Inference Attacks under Data Dependencies\",\"authors\":\"Thomas Humphries, Simon Oya, Lindsey Tulloch, Matthew Rafuse, I. Goldberg, U. Hengartner, F. Kerschbaum\",\"doi\":\"10.1109/CSF57540.2023.00013\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Training machine learning models on privacy-sensitive data has become a popular practice, driving innovation in ever-expanding fields. This has opened the door to new attacks that can have serious privacy implications. One such attack, the Membership Inference Attack (MIA), exposes whether or not a particular data point was used to train a model. A growing body of literature uses Differentially Private (DP) training algorithms as a defence against such attacks. However, these works evaluate the defence under the restrictive assumption that all members of the training set, as well as non-members, are independent and identically distributed. This assumption does not hold for many real-world use cases in the literature. Motivated by this, we evaluate membership inference with statistical dependencies among samples and explain why DP does not provide meaningful protection (the privacy parameter $\\\\epsilon$ scales with the training set size $n$) in this more general case. We conduct a series of empirical evaluations with off-the-shelf MIAs using training sets built from real-world data showing different types of dependencies among samples. Our results reveal that training set dependencies can severely increase the performance of MIAs, and therefore assuming that data samples are statistically independent can significantly underestimate the performance of MIAs.\",\"PeriodicalId\":179870,\"journal\":{\"name\":\"2023 IEEE 36th Computer Security Foundations Symposium (CSF)\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2020-10-23\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"12\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2023 IEEE 36th Computer Security Foundations Symposium (CSF)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/CSF57540.2023.00013\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2023 IEEE 36th Computer Security Foundations Symposium (CSF)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/CSF57540.2023.00013","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 12

摘要

在隐私敏感数据上训练机器学习模型已经成为一种流行的做法,推动了不断扩大的领域的创新。这为可能造成严重隐私问题的新攻击打开了大门。其中一种攻击是成员推理攻击(MIA),它暴露了是否使用了特定的数据点来训练模型。越来越多的文献使用差分私有(DP)训练算法来防御此类攻击。然而,这些工作在限制性假设下评估防御,即训练集的所有成员以及非成员都是独立且均匀分布的。在文献中,这个假设并不适用于许多真实世界的用例。受此启发,我们用样本之间的统计依赖性来评估隶属性推断,并解释为什么DP在这种更一般的情况下不能提供有意义的保护(隐私参数$\epsilon$与训练集大小$n$相对应)。我们使用从真实世界数据构建的训练集对现成的MIAs进行了一系列的经验评估,这些数据显示了样本之间不同类型的依赖关系。我们的研究结果表明,训练集依赖性可以严重提高MIAs的性能,因此假设数据样本在统计上是独立的可以显著低估MIAs的性能。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
Investigating Membership Inference Attacks under Data Dependencies
Training machine learning models on privacy-sensitive data has become a popular practice, driving innovation in ever-expanding fields. This has opened the door to new attacks that can have serious privacy implications. One such attack, the Membership Inference Attack (MIA), exposes whether or not a particular data point was used to train a model. A growing body of literature uses Differentially Private (DP) training algorithms as a defence against such attacks. However, these works evaluate the defence under the restrictive assumption that all members of the training set, as well as non-members, are independent and identically distributed. This assumption does not hold for many real-world use cases in the literature. Motivated by this, we evaluate membership inference with statistical dependencies among samples and explain why DP does not provide meaningful protection (the privacy parameter $\epsilon$ scales with the training set size $n$) in this more general case. We conduct a series of empirical evaluations with off-the-shelf MIAs using training sets built from real-world data showing different types of dependencies among samples. Our results reveal that training set dependencies can severely increase the performance of MIAs, and therefore assuming that data samples are statistically independent can significantly underestimate the performance of MIAs.
求助全文
通过发布文献求助,成功后即可免费获取论文全文。 去求助
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信