arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2011.11877cs.LGcs.CCcs.CRcs.DSstat.ML
收藏

InstaHide在混合两张私有图像时的样本复杂度

InstaHide's Sample Complexity When Mixing Two Private Images

University of California, Berkeley(加州大学伯克利分校);Adobe Research(奥多比研究院);Columbia University(哥伦比亚大学);Boston University(波士顿大学);Simons Institute for the Theory of Computing(西蒙斯理论计算研究所);Duke University(杜克大学)

Baihe Huang, Zhao Song, Runzhou Tao, Junze Yin, Ruizhe Zhang, Danyang Zhuo

AI总结:

本文针对InstaHide隐私保护方案,提出统一攻击分析框架,并设计新算法在最优样本复杂度下可证明恢复所有私有图像,同时证明其计算安全性。

摘要:

训练神经网络通常需要大量的敏感训练数据,因此如何保护训练数据的隐私已成为深度学习研究中的一个关键课题。InstaHide是一种最先进的方案,旨在保护训练数据隐私,同时对测试准确率影响较小,其安全性已成为一个突出问题。在本文中,我们系统地研究了近期针对InstaHide的攻击,并提出了一个统一框架来理解和分析这些攻击。我们发现现有攻击要么没有可证明的保证,要么只能恢复单个私有图像。在当前InstaHide挑战设置下,每个InstaHide图像是两张私有图像的混合,我们提出了一种新算法,能够在可证明的保证和最优样本复杂度下恢复所有私有图像。此外,我们还提供了检索所有InstaHide图像的计算困难性结果。我们的结果表明,即使在混合两张私有图像的情况下,InstaHide在信息论上不是安全的,但在最坏情况下是计算上安全的。

英文摘要:

Training neural networks usually require large numbers of sensitive training data, and how to protect the privacy of training data has thus become a critical topic in deep learning research. InstaHide is a state-of-the-art scheme to protect training data privacy with only minor effects on test accuracy, and its security has become a salient question. In this paper, we systematically study recent attacks on InstaHide and present a unified framework to understand and analyze these attacks. We find that existing attacks either do not have a provable guarantee or can only recover a single private image. On the current InstaHide challenge setup, where each InstaHide image is a mixture of two private images, we present a new algorithm to recover all the private images with a provable guarantee and optimal sample complexity. In addition, we also provide a computational hardness result on retrieving all InstaHide images. Our results demonstrate that InstaHide is not information-theoretically secure but computationally secure in the worst case, even when mixing two private images.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑