MadryLab / cifar10_challenge

A challenge to explore adversarial robustness of neural networks on CIFAR10.

MadryLab/cifar10_challenge Issues

pretrained model link expired
Closed 3 years ago1
Submitting my result to white CIFAR-10 leaderboards
Closed 4 years ago1
lack of Sign function
Closed 4 years ago
The config of training robust model for CIFAR10.
Closed 4 years ago1
pytorch definition of the model
Closed 4 years ago1
clean accuracy on adversarially trained cifar10 resnet 18
Closed 5 years ago1
Naturally trained network gives 78% test accuracy on pgd attack.
Closed 5 years ago1
About the network architecture
Closed 5 years ago1
Making adversarial examples during training
Closed 5 years ago1
Overflow when random_restart is false
Closed 5 years ago1
Number of trainable parameters
Closed 5 years ago1
When generating uniform noise in random start, floating point number will cause invalid pixel value.
Closed 5 years ago1
Matching training statistics
Closed 5 years ago2
PGD steps along the sign of the gradient
Closed 5 years ago1
About the convergence of training.
Closed 5 years ago2
Dataset normalization
Closed 5 years ago2
Same accuracy logged twice
Closed 5 years ago1
Base network questions and implementation.
Closed 5 years ago3
Googlenet with owndata
Closed 5 years ago1
Image Channels
Closed 5 years ago1
About the accuracy of adversarial examples
Closed 5 years ago1
about the loss in the pgd_attack
Closed 6 years ago1
White-box result of madry_lab_challenges in examples of cleverhans.
Closed 6 years ago1
config for 20 iteration pgd that gives 47.04% accuracy.
Closed 6 years ago1
Questions about recreating paper results
Closed 6 years ago4
How to determine "best" model
Closed 6 years ago4
Image out of valid range for the first iteration of PGD attack
Closed 6 years ago2
cifar_input.py Function get_next_batch() has a small bug.
Closed 6 years ago1