attack.LabelConsistent

class LabelConsistent[source]

Bases: BadNet

Label-Consistent Backdoor Attacks

link : https://github.com/MadryLab/label-consistent-backdoor-code

basic structure:

  1. config args, save_path, fix random seed

  2. set the clean train data and clean test data

  3. set the attack img transform and label transform

  4. set the backdoor attack data and backdoor test data

  5. set the device, model, criterion, optimizer, training schedule.

  6. attack or use the model to do finetune with 5% clean data

  7. save the attack result for defense

attack = LabelConsistent()
attack.attack()

Note

@article{turner2019labelconsistent, title = {Label-Consistent Backdoor Attacks}, author = {Alexander Turner and Dimitris Tsipras and Aleksander Madry}, journal = {arXiv preprint arXiv:1912.02771}, year = {2019}}

Parameters:

  • attack (string) – name of attack, use to match the transform and set the saving prefix of path.

  • attack_target (Int) – target class No. in all2one attack

  • attack_label_trans (str) – which type of label modification in backdoor attack

  • pratio (float) – the poison rate

  • bd_yaml_path (string) – path for yaml file provide additional default attributes

  • attack_train_replace_imgs_path (string) – path for adversarial-attacked images, since we need images to be adversarial attacked and then we add patch trigger onto them. If not provided, we will use the default path.

  • reduced_amplitude (float) – the alpha/transparency of the backdoor trigger added at corners

  • resource_folder_path (string) – where the resource folder is

  • **kwargs (optional) – Additional attributes.