AI-security-homework

"AI Security" coursework. (The files content are shown below：)

Task_1: Train Alexnet on mnist.
Task_2: White box attack for classification.(FGSM, IFGSM, and MIFGSM)
Task_3: Adv-GAN(Test Black-box transfer attack on torchvision.models.xxxxxx)
Task_4: Adversarial defense by pre-processing(JPEG, TVM, Image Quilting, and MagNet).
Task_5: Adversarial defense by adversarial training.(FreeAT)
Task_6: Black-box attack in self-use scenarios.