BppAttack: Stealthy and Efficient Trojan Attacks against Deep Neural Networks via Image Quantization and Contrastive Adversarial Learning. [pdf] [code] Zhenting Wang, Juan Zhai, and Shiqing Ma. CVPR, 2022. Dynamic Backdoor Attacks Against Machine Learning Models. [pdf] Ahmed Salem, Rui Wen, Mich...
除了我们对基本的和复杂的后门的防御效果之外,我们的论文的一个意想不到的特点是两种后门注射方法之间的显著差异:触发器驱动的具有完全访问模型训练的Badnet端对端攻击和神经元驱动的Trojan Attack,无需访问模型训练。然而,权衡通过实验,我们发现Trojan Attack注入方法通常会增加不必要的扰动,并给非目标神经元带来不可预...
via Decoupling the Training Process}, author={Huang, Kunzhe and Li, Yiming and Wu, Baoyuan and Qin, Zhan and Ren, Kui}, booktitle={International Conference on Learning Representations}, year={2022} } @inproceedings{ssba-backdoor-attack-iccv2021, title={Invisible backdoor attack with sample-...