arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2210.13631cs.LGcs.CR
收藏

数据集推理的鲁棒性研究

On the Robustness of Dataset Inference

Aalto University(阿尔托大学);Zhejiang University(浙江大学);University of Waterloo(滑铁卢大学)

Sebastian Szyller, Rui Zhang, Jian Liu, N. Asokan

AI总结:

本研究证明数据集推理(DI)在所有权验证中存在高误报和漏报问题,黑盒设置下无法可靠识别被窃模型,并探讨了其影响与未来方向。

摘要:

机器学习(ML)模型训练成本高昂,因为需要大量数据、计算资源和技术专长。因此,它们构成有价值的知识产权,需要保护免受试图窃取它们的对手的侵害。所有权验证技术允许模型窃取攻击的受害者证明可疑模型实际上是从他们的模型窃取的。尽管已经提出了许多基于水印或指纹识别的所有权验证技术,但大多数在安全保证(装备精良的对手可以逃避验证)或计算成本方面存在不足。一种指纹识别技术——数据集推理(DI)已被证明比先前方法提供更好的鲁棒性和效率。DI的作者为线性(可疑)模型提供了正确性证明。然而,在同一设置的子空间中,我们证明DI遭受高误报率(FPs)——它可能错误地将使用来自同一分布的非重叠数据训练的独立模型识别为被窃取的模型。我们进一步证明DI在现实的非线性可疑模型中也会触发误报。然后,我们通过实验证实,在黑盒设置中,DI会导致高置信度的误报。其次,我们表明DI也遭受漏报(FNs)——对手可以通过对抗训练正则化被窃取模型的决策边界来欺骗DI(以牺牲一定准确性为代价),从而导致漏报。为此,我们证明黑盒DI无法识别从被窃取数据集进行对抗训练的模型——这是DI最难逃避的设置。最后,我们讨论了我们的发现的影响、基于指纹识别的所有权验证的一般可行性,并提出了未来工作的方向。

英文摘要:

Machine learning (ML) models are costly to train as they can require a significant amount of data, computational resources and technical expertise. Thus, they constitute valuable intellectual property that needs protection from adversaries wanting to steal them. Ownership verification techniques allow the victims of model stealing attacks to demonstrate that a suspect model was in fact stolen from theirs. Although a number of ownership verification techniques based on watermarking or fingerprinting have been proposed, most of them fall short either in terms of security guarantees (well-equipped adversaries can evade verification) or computational cost. A fingerprinting technique, Dataset Inference (DI), has been shown to offer better robustness and efficiency than prior methods. The authors of DI provided a correctness proof for linear (suspect) models. However, in a subspace of the same setting, we prove that DI suffers from high false positives (FPs) -- it can incorrectly identify an independent model trained with non-overlapping data from the same distribution as stolen. We further prove that DI also triggers FPs in realistic, non-linear suspect models. We then confirm empirically that DI in the black-box setting leads to FPs, with high confidence. Second, we show that DI also suffers from false negatives (FNs) -- an adversary can fool DI (at the cost of incurring some accuracy loss) by regularising a stolen model's decision boundaries using adversarial training, thereby leading to an FN. To this end, we demonstrate that black-box DI fails to identify a model adversarially trained from a stolen dataset -- the setting where DI is the hardest to evade. Finally, we discuss the implications of our findings, the viability of fingerprinting-based ownership verification in general, and suggest directions for future work.

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。

↑