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 2403.15918cs.CV
收藏

面向自监督学习(SSL)中的对抗鲁棒性与后门缓解

Towards Adversarial Robustness And Backdoor Mitigation in SSL

Centre for Development of Advanced Computing(高级计算开发中心)

Aryan Satpathy, Nilaksh Singh, Dhruva Rajwade, Somesh Kumar

AI总结:

本研究针对自监督学习(SSL)易受后门攻击的问题,提出计算高效且泛用的防御方法,结合频域增强提升对抗鲁棒性,在多基准上验证其可缓解后门攻击并维持下游任务性能。

摘要:

自监督学习(Self-Supervised Learning,SSL)在从无标注数据中学习表征方面展现出巨大潜力。无需人工标注即可学习表征的优势使SSL成为现实问题中广泛应用的技术。然而,近期研究表明SSL方法易受后门攻击,攻击者可通过篡改训练数据分布或修改模型本身,利用所学模型操纵学习到的表征。本研究旨在解决SSL中的后门攻击防御问题,攻击场景为攻击者可获取SSL训练数据的真实占比子集,但无法访问模型。我们提出了计算高效且能在不同问题场景下泛化的新方法。我们还研究了经本方法训练的SSL模型的对抗鲁棒性,并揭示了通过频域增强提升SSL鲁棒性的机制。我们在多种SSL基准上验证了方法的有效性,结果表明该方法能够缓解后门攻击,同时在下游任务上保持高性能。本研究代码可在github.com/Aryan-Satpathy/Backdoor获取。

英文摘要:

Self-Supervised Learning (SSL) has shown great promise in learning representations from unlabeled data. The power of learning representations without the need for human annotations has made SSL a widely used technique in real-world problems. However, SSL methods have recently been shown to be vulnerable to backdoor attacks, where the learned model can be exploited by adversaries to manipulate the learned representations, either through tampering the training data distribution, or via modifying the model itself. This work aims to address defending against backdoor attacks in SSL, where the adversary has access to a realistic fraction of the SSL training data, and no access to the model. We use novel methods that are computationally efficient as well as generalizable across different problem settings. We also investigate the adversarial robustness of SSL models when trained with our method, and show insights into increased robustness in SSL via frequency domain augmentations. We demonstrate the effectiveness of our method on a variety of SSL benchmarks, and show that our method is able to mitigate backdoor attacks while maintaining high performance on downstream tasks. Code for our work is available at github.com/Aryan-Satpathy/Backdoor

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。

↑