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 2302.01375cs.LGcs.AI
收藏

随机集成对对抗扰动的鲁棒性研究

On the Robustness of Randomized Ensembles to Adversarial Perturbations

University of Illinois at Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校)

Hassan Dbouk, Naresh R. Shanbhag

AI总结:

本文从理论上分析了随机集成分类器(RECs)的极限与有用条件,并提出新的boosting算法BARRE,实验证明其能有效防御强ℓ∞范数对抗攻击。

摘要:

随机集成分类器(RECs)在推理时随机选择一个分类器,已成为传统集成方法的一种有吸引力的替代方案,用于在有限计算需求下实现对抗鲁棒分类器。然而,近期研究表明,现有构建RECs的方法比最初声称的更为脆弱,这对其有效性产生了重大质疑,并引发了根本性问题,例如:“RECs何时有用?”、“它们的局限性是什么?”以及“我们如何训练它们?”。在这项工作中,我们首先揭开RECs的神秘面纱,推导出关于其理论极限、其有用性的必要和充分条件等基本结果。利用这一新认识,我们提出了一种新的boosting算法(BARRE)用于训练鲁棒的RECs,并通过实验证明其在各种网络架构和数据集上防御强ℓ∞范数有界对手的有效性。我们的代码可在https://github.com/hsndbk4/BARRE找到。

英文摘要:

Randomized ensemble classifiers (RECs), where one classifier is randomly selected during inference, have emerged as an attractive alternative to traditional ensembling methods for realizing adversarially robust classifiers with limited compute requirements. However, recent works have shown that existing methods for constructing RECs are more vulnerable than initially claimed, casting major doubts on their efficacy and prompting fundamental questions such as: "When are RECs useful?", "What are their limits?", and "How do we train them?". In this work, we first demystify RECs as we derive fundamental results regarding their theoretical limits, necessary and sufficient conditions for them to be useful, and more. Leveraging this new understanding, we propose a new boosting algorithm (BARRE) for training robust RECs, and empirically demonstrate its effectiveness at defending against strong $\ell_\infty$ norm-bounded adversaries across various network architectures and datasets. Our code can be found at https://github.com/hsndbk4/BARRE.

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。

↑