Code for "Transferable Availability Poisoning Attacks"
python anneal.py --net ResNet18 --dataset CIFAR10 --recipe targeted --eps 8 --budget 1.0 --save poison_dataset --poison_path ./results/resnet18_cifar10_tp --attackoptim PGD --cl_alg SimCLR --allow_mmt_grad --restarts 1
- Eval with supervised learning
python poison_evaluation/main.py --load_path ./results/resnet18_cifar10_tp/ --runs 1
- Eval with supervised contrastive learning
python main.py --dataset tpcifar10 --arch resnet18 --cl_alg SupCL --folder_name baseline_tp --baseline --epochs 1000 --eval_freq 100
- Eval with semi-supervised learning (FixMatch)
python FixMatch/Train_fixmatch.py --seed 1 --arch resnet18 --dataset tpcifar10 --n_classes 10
- Eval with SimSiam
python main.py --dataset tpcifar10 --arch resnet18 --cl_alg SimSiam --folder_name baseline_tp --baseline --epochs 1000 --eval_freq 100
- Eval with SimCLR
python main.py --dataset tpcifar10 --arch resnet18 --cl_alg SimCLR --folder_name baseline_tp --baseline --epochs 1000 --eval_freq 100