huggingface/pytorch-image-models

AdvProp implementation

Offen

#68 geöffnet am 03.01.2020

 (10 Kommentare) (1 Reaktion) (0 zugewiesene Personen)Python (4.595 Forks)batch import
enhancementhelp wanted

Repository-Metriken

Stars
 (30.131 Sterne)
PR-Merge-Metriken
 (Durchschn. Merge 6T 7h) (4 gemergte PRs in 30 T)

Beschreibung

https://arxiv.org/abs/1911.09665

In the paper, they propose calculating two losses: one for the forward pass with "clean" BN params, and another for the forward pass with adversarial BN params. Then they combine these two losses, and backprop through both BN paths at the same time (joint optimization).

Does the following look correct to you:

class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.conv1 = nn.Conv2d(3, 32, kernel_size=5, stride=2)
        self.bnC = nn.BatchNorm2d(32)
        self.bnA = nn.BatchNorm2d(32)
        self.relu = nn.ReLU()
        self.linear = nn.Linear(32*14*14, 10)

    def forward(self, x, clean=True):
        x = self.conv(x)
        if clean:
            x = self.bnC(x)
        else:
            x = self.bnA(x)
        x = self.relu(x)
        x = self.linear(x)
        return x

model = Net()
optimizer = torch.optim.SGD(model.parameters(), lr=0.01)
loss_fn = nn.CrossEntropyLoss()

for i in range(1000):
    batchC, targetC = get_clean_batch()
    batchA, targetA = get_adv_batch()

    outputC = model(batchC, clean=True)
    outputA = model(batchA, clean=False)

    lossC = loss_fn(outputC, targetC)
    lossA = loss_fn(outputA, targetA)
    loss = lossC + lossA

    optimizer.zero_grad()
    loss.backward()
    optimizer.step()

If so, how would you propagate clean argument to all the blocks, especially the ones that use nn.Sequential lists?

Is there some existing AdvProp code to look at?

Contributor Guide