arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2401.09574cs.LGcs.CR
收藏

迈向可扩展且鲁棒的模型版本控制

Towards Scalable and Robust Model Versioning

University of Chicago(芝加哥大学)

Wenxin Ding, Arjun Nitin Bhagoji, Ben Y. Zhao, Haitao Zheng

AI总结:

针对深度学习模型部署后易遭白盒对抗攻击的问题,本文提出无需新数据和架构改动的参数化隐藏分布模型版本控制方法,可生成多版本抵御泄露版本的攻击,显著提升DNN分类器鲁棒性。

摘要:

随着深度学习模型在各行业的部署规模持续扩大,旨在获取已部署模型访问权限的恶意入侵威胁正不断上升。一旦攻击者通过服务器入侵、内部攻击或模型反演技术获取了已部署模型的访问权限,他们就能构建白盒对抗性攻击来操纵模型的分类结果,从而给依赖这些模型执行关键任务的组织带来重大风险。模型所有者需要相关机制来防范此类损失,且无需获取新的训练数据——获取新训练数据的过程通常需要投入大量时间与资金。\n本文探讨了在不获取新训练数据、不改变模型架构的前提下,生成具备不同攻击特性的多个模型版本的可行性。模型所有者可一次部署一个版本,当某个版本泄露时,立即用新版本替换。新部署的模型版本能够抵御利用一个或所有先前泄露版本的白盒访问权限生成的对抗性攻击。我们从理论上证明,通过在模型训练数据中引入参数化隐藏分布,迫使模型学习由所选数据唯一决定的任务无关特征,即可实现这一目标。此外,对隐藏分布的最优选择能够生成一系列模型版本,可随时间推移抵御复合可迁移性攻击。基于我们的分析结论,我们为DNN(深度神经网络)分类器设计并实现了一种实用的模型版本控制方法,该方法相比现有方法显著提升了鲁棒性。我们认为,本研究为DNN服务在初始部署后的安全防护提供了一个颇具前景的方向。

英文摘要:

As the deployment of deep learning models continues to expand across industries, the threat of malicious incursions aimed at gaining access to these deployed models is on the rise. Should an attacker gain access to a deployed model, whether through server breaches, insider attacks, or model inversion techniques, they can then construct white-box adversarial attacks to manipulate the model's classification outcomes, thereby posing significant risks to organizations that rely on these models for critical tasks. Model owners need mechanisms to protect themselves against such losses without the necessity of acquiring fresh training data - a process that typically demands substantial investments in time and capital. In this paper, we explore the feasibility of generating multiple versions of a model that possess different attack properties, without acquiring new training data or changing model architecture. The model owner can deploy one version at a time and replace a leaked version immediately with a new version. The newly deployed model version can resist adversarial attacks generated leveraging white-box access to one or all previously leaked versions. We show theoretically that this can be accomplished by incorporating parameterized hidden distributions into the model training data, forcing the model to learn task-irrelevant features uniquely defined by the chosen data. Additionally, optimal choices of hidden distributions can produce a sequence of model versions capable of resisting compound transferability attacks over time. Leveraging our analytical insights, we design and implement a practical model versioning method for DNN classifiers, which leads to significant robustness improvements over existing methods. We believe our work presents a promising direction for safeguarding DNN services beyond their initial deployment.

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。

↑