Steganografiya quyidagi sohalarda qo'llaniladi, lekin ular bilan cheklanmaydi

item.tobytes(),
hashlib.sha256).hexdigest()
bits = bitstring.BitArray(hex=hashed).bin
if int(bits[:self.args.precision_dawn], 2) <= bound:
return False
else:
return True

def generate_trigger_dawn(self):
"""Generation trigger set based on the paper.

DAWN: Dynamic Adversarial Watermarking of Neural Networks

by Szyller et al.

Returns:
trainloader (Object): training loader
valloader (Object): validation loader
testloader (Object): test loader
triggerloader (Object): trigger loader

"""

trainloader = torch.utils.data.DataLoader(
self.trainset, batch_size=1, shuffle=True)
triggerset = []
for _, data in enumerate(trainloader):
inputs, labels = data
shapes = list(inputs.size())
if not self.__is_prediction_dawn(inputs.numpy()):
triggerset.append((inputs.reshape(shapes[1:]), labels))

triggerloader = torch.utils.data.DataLoader(
triggerset, batch_size=100, shuffle=True)

return None, None, None, triggerloader

def train_step(self, X, Y, idx):
"""Training step

Args:
X (Tensor): input data
Y (Tensor): label data
idx (int): batch id

"""
# Compute loss for original data
inputs, labels = X.to(self.device), Y.to(self.device)
self.optimizer.zero_grad()
outputs = self.model(inputs)
loss = self.args.criterion(outputs, labels)
# Watermark loss if required
if self.watermark:
if idx % self.args.interval_wm == 0:
wmloss = self.watermark_loss()
loss += wmloss