在能力受限情况下发起强大的后门攻击（英文版）

英文标题：Launching a Robust Backdoor Attack under Capability Constrained Scenarios中文摘要：本研究探讨了在能力限制内实施黑盒后门攻击的可能性，通过设计后门触发器，攻击者可以在没有参与训练过程或了解目标模型结构的情况下作为图像注释者或供应商从事此类攻击。实验结果表明，我们的方法在黑盒场景中实现了高攻击成功率，并逃过了最先进的后门防御。英文摘要：As deep neural networks continue to be used in critical domains, concernsover their security have emerged. Deep learning models are vulnerable tobackdoor attacks due to the lack of transparency. A poisoned backdoor model mayperform normally in routine environments, but exhibit malicious be