CVPR2021-Competition-Unrestricted-Adversarial-Attacks-on-ImageNet
CVPR2021-Competition-Unrestricted-Adversarial-Attacks-on-ImageNet copied to clipboard
关于pytorch框架下如何攻击集成对抗训练的模型
您好,不好意思,打扰您了。我有一些关于对抗攻击的问题想请教一下您。 在一些关于tranfer-based attack的论文中经常可以看到文中攻击的对抗训练防御的模型为adversarially trained Inception-v3 (Inc-v3adv), ensemble of three adversarially trained Inception-v3 models (Inc-v3ens3), ensemble of four adversarially trained Inception-v3 models (Inc-v3ens4), adversarially trained Inception-Resnet-v2 (IncRes-v2adv) and ensemble of three adversarially trained Inception-Resnet-v2 models (IncRes-v2ens3),这些经过对抗训练的模型都是tensorflow框架下的模型,我想问一下在pytorch框架下有公开的类似于上述模型的对抗训练模型吗?
好的,十分感谢您的回复!之前拜读过您的一些论文,实验效果很好,对我也很有启发。再次感谢您!