

The code is also available at github.com/advadvadvadvadv/adv, along with weights for trained ResNet110 models on CIFAR10

1. Seen attacks

1.a without regularization

```
python3 adv26b.py -dat cifar10 -mod resnet110 -att wbf -lmt 0 -tau 0 -bas 100 -vls 0.9 -tss 0.1 -see 0
```

1.b with regularization

```
python3 adv26b.py -dat cifar10 -mod resnet110 -att wbf -lmt 1 -tau 1 -bas 100 -vls 0.9 -tss 0.1 -see 0
```

2. Unseen attacks

2.a without regularization

```
python3 adv26b-gen.py -dat cifar10 -mod resnet110 -att wb -lmt 0 -tau 0 -bas 100 -vls 0.9 -tss 0.1 -see 0
```

2.b with regularization

```
python3 adv26b-gen.py -dat cifar10 -mod resnet110 -att wb -lmt 1 -tau 1 -bas 100 -vls 0.9 -tss 0.1 -see 0
```
