Steganografiya quyidagi sohalarda qo'llaniladi, lekin ular bilan cheklanmaydi

elif isinstance(self.model, RandomForestRegressor):
self.model.fit(X_train, y_train)
predictions = self.model.predict(ownership['inputs'])
for q in [1, 0.5, 0.3, 0.25, 0.2, 0.1, 0.05, 0.02]:
selected_q = number_labels * q
if verify(
ownership['labels'],
predictions,
number_labels=selected_q,
bounds=ownership['bounds'],
metric=self.metric)['is_stolen']:

break

return selected_q

else:
raise NotImplementedError()

def __is_prediction_dawn(self, item):
""" Verify if prediction should be
correct as in the DAWN paper.

Args:
item (array): Single query

Returns:
(bool): Should the input be correctly classified ?

"""
bound = floor((2 ** self.args.precision_dawn)
* self.args.probability_dawn)
hashed = hmac.new(
self.args.key_dawn.encode("utf-8"),
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 predict(self, X_test):
"""Predict method.

Args:
X_test (array): Test dataset

Returns:
predictions (array): Predictions

"""
# Verify if watermark is DAWN
if self.args.trigger_technique == 'dawn':
true_results = self.model.predict(X_test)
returned_results = []
# Loop on all queries
for x_item, x_true in zip(X_test, true_results):
if self.__is_prediction_dawn(x_item):
returned_results.append(x_true)