arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2104.10706stat.MLcs.CRcs.LG
收藏

数据集推理:机器学习中的所有权解析

Dataset Inference: Ownership Resolution in Machine Learning

IIT Delhi(印度理工学院德里分校);University of Toronto(多伦多大学);Vector Institute(向量研究所)

Pratyush Maini, Mohammad Yaghini, Nicolas Papernot

摘要:

随着训练中涉及越来越多的数据和计算,机器学习模型构成了宝贵的知识产权。这激发了人们对模型窃取的兴趣,通过在部分、很少或没有监督的情况下学习的进步,这变得更加实用。现有的防御集中在模型的决策面中插入独特的水印,但这是不够的:水印不是从训练分布中采样的,因此在模型窃取期间并不总是保留。在本文中,我们进行了关键的观察,即被盗模型的训练集中包含的知识是所有被盗副本所共有的。对手的目标,无论采用的攻击,总是提取这种知识或其副产品。这给了原始模型的所有者一个强大的优势:模型所有者可以访问原始训练数据。因此,我们引入$dataset$ $inference$,识别可疑模型副本是否具有来自原始模型数据集的私有知识的过程,作为对模型窃取的防御。我们开发了一种数据集推理方法,该方法将统计测试与估计多个数据点到决策边界的距离的能力相结合。我们在CIFAR10、SVHN、CIFAR100和ImageNet上的实验表明,模型所有者可以以超过99%的信心声称他们的模型(或数据集)被盗,尽管只暴露了被盗模型的50个训练点。数据集推理可以抵御最先进的攻击,即使对手是自适应的。与以前的工作不同,它不需要重新训练或过度拟合被辩护的模型。

英文摘要:

With increasingly more data and computation involved in their training, machine learning models constitute valuable intellectual property. This has spurred interest in model stealing, which is made more practical by advances in learning with partial, little, or no supervision. Existing defenses focus on inserting unique watermarks in a model's decision surface, but this is insufficient: the watermarks are not sampled from the training distribution and thus are not always preserved during model stealing. In this paper, we make the key observation that knowledge contained in the stolen model's training set is what is common to all stolen copies. The adversary's goal, irrespective of the attack employed, is always to extract this knowledge or its by-products. This gives the original model's owner a strong advantage over the adversary: model owners have access to the original training data. We thus introduce $dataset$ $inference$, the process of identifying whether a suspected model copy has private knowledge from the original model's dataset, as a defense against model stealing. We develop an approach for dataset inference that combines statistical testing with the ability to estimate the distance of multiple data points to the decision boundary. Our experiments on CIFAR10, SVHN, CIFAR100 and ImageNet show that model owners can claim with confidence greater than 99% that their model (or dataset as a matter of fact) was stolen, despite only exposing 50 of the stolen model's training points. Dataset inference defends against state-of-the-art attacks even when the adversary is adaptive. Unlike prior work, it does not require retraining or overfitting the defended model.

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。

↑