arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2107.05166cs.LGcs.AIcs.CRstat.ML
收藏

模型窃取攻击的状态检测

Stateful Detection of Model Extraction Attacks

Indian Institute of Science(印度科学学院);nference(nference公司)

Soham Pal, Yash Gupta, Aditya Kanade, Shirish Shevade

摘要:

机器学习即服务提供商通过应用程序编程接口(API)向开发人员公开机器学习(ML)模型。最近的工作表明,攻击者可以利用这些API来提取这些ML模型的良好近似,通过使用他们选择的样本查询它们。我们提出VarDetect,一个有状态的监视器,跟踪这样的服务的用户所做的查询的分布,检测模型窃取攻击。利用修改后的变分自编码器学习的潜在分布,VarDetect鲁棒地将三种类型的攻击者样本与良性样本分开,并成功地为每种样本发出警报。此外,随着VarDetect部署为自动防御机制,发现提取的替代模型表现出预期的性能和可移植性较差。最后,我们证明,即使是自适应攻击者与部署VarDetect的先验知识,检测。

英文摘要:

Machine-Learning-as-a-Service providers expose machine learning (ML) models through application programming interfaces (APIs) to developers. Recent work has shown that attackers can exploit these APIs to extract good approximations of such ML models, by querying them with samples of their choosing. We propose VarDetect, a stateful monitor that tracks the distribution of queries made by users of such a service, to detect model extraction attacks. Harnessing the latent distributions learned by a modified variational autoencoder, VarDetect robustly separates three types of attacker samples from benign samples, and successfully raises an alarm for each. Further, with VarDetect deployed as an automated defense mechanism, the extracted substitute models are found to exhibit poor performance and transferability, as intended. Finally, we demonstrate that even adaptive attackers with prior knowledge of the deployment of VarDetect, are detected by it.

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。

↑