arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2302.00747cs.LG
收藏

Universal Soldier:利用通用对抗扰动检测后门攻击

Universal Soldier: Using Universal Adversarial Perturbations for Detecting Backdoor Attacks

Radboud University(拉德堡德大学)

Xiaoyun Xu, Oguzhan Ersoy, Stjepan Picek

AI总结:

针对无触发器先验时难区分干净与带后门模型的问题,提出基于通用对抗扰动(UAP)的USB后门检测方法,利用带后门模型UAP扰动量更小的特性检测后门并逆向触发器,在345个模型上效果达SOTA水平。

摘要:

深度学习模型在众多机器学习任务中表现优异,但也面临对抗样本、投毒(后门)攻击等安全问题。深度学习模型可能因使用带后门的数据训练、或内部网络参数被篡改而遭投毒。此时,带后门的模型接收干净输入时表现符合预期,但接收印有被称为“触发器”的预设模式的带后门输入时会发生误分类。遗憾的是,在不掌握触发器先验知识的情况下,很难区分干净模型与带后门模型。本文提出一种后门检测方法,该方法利用一类特殊的对抗攻击——通用对抗扰动(universal adversarial perturbation, UAP),及其与后门触发器的相似性。我们观察到一个直观现象:从带后门模型生成的UAP误导模型所需的扰动量,少于从干净模型生成的UAP。带后门模型的UAP倾向于利用由后门触发器构建的、从所有类别到目标类别的捷径。我们提出一种名为Universal Soldier for Backdoor detection(USB,后门检测通用方法)的新方法,并通过UAP逆向工程潜在的后门触发器。在多个数据集上训练的345个模型上开展的实验表明,USB可有效检测注入的后门,其效果与现有最优方法相当或更优。

英文摘要:

Deep learning models achieve excellent performance in numerous machine learning tasks. Yet, they suffer from security-related issues such as adversarial examples and poisoning (backdoor) attacks. A deep learning model may be poisoned by training with backdoored data or by modifying inner network parameters. Then, a backdoored model performs as expected when receiving a clean input, but it misclassifies when receiving a backdoored input stamped with a pre-designed pattern called "trigger". Unfortunately, it is difficult to distinguish between clean and backdoored models without prior knowledge of the trigger. This paper proposes a backdoor detection method by utilizing a special type of adversarial attack, universal adversarial perturbation (UAP), and its similarities with a backdoor trigger. We observe an intuitive phenomenon: UAPs generated from backdoored models need fewer perturbations to mislead the model than UAPs from clean models. UAPs of backdoored models tend to exploit the shortcut from all classes to the target class, built by the backdoor trigger. We propose a novel method called Universal Soldier for Backdoor detection (USB) and reverse engineering potential backdoor triggers via UAPs. Experiments on 345 models trained on several datasets show that USB effectively detects the injected backdoor and provides comparable or better results than state-of-the-art methods.

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。

↑