Fix documentation

Signed-off-by: Beat Buesser <[email protected]>

Author: Beat Buesser

art/attacks/evasion/pe_malware_attack.py

@@ -30,13 +30,14 @@
 from tqdm.auto import trange
 
 import numpy as np
-import tensorflow as tf
 
 from art.estimators.estimator import BaseEstimator, NeuralNetworkMixin
 from art.estimators.classification.classifier import ClassifierMixin
 from art.attacks.attack import EvasionAttack
 
 if TYPE_CHECKING:
+    # pylint: disable=C0412
+    import tensorflow as tf
     from art.utils import CLASSIFIER_NEURALNETWORK_TYPE
 
 logger = logging.getLogger(__name__)
@@ -212,6 +213,8 @@ def generate_mask(
                                of perturbation region.
         :return mask: Array with 1s on the features we will modify on this batch and 0s elsewhere.
         """
+        import tensorflow as tf  # lgtm [py/repeated-import]
+
         mask = np.zeros_like(x)
         for i in range(len(x)):
             if y[i] == 1:
@@ -252,6 +255,8 @@ def update_embeddings(self, embeddings: "tf.Tensor", gradients: "tf.Tensor", mas
         :param mask: Tensor with 1s on the embeddings we modify, 0s elsewhere.
         :return embeddings: Updated embeddings wrt the adversarial objective.
         """
+        import tensorflow as tf  # lgtm [py/repeated-import]
+
         if self.use_sign:
             gradients = tf.sign(gradients)
         embeddings = embeddings + self.l_r * gradients * mask
@@ -282,6 +287,8 @@ def get_adv_malware(
 
         :return data: Numpy array with valid data samples.
         """
+        import tensorflow as tf  # lgtm [py/repeated-import]
+
         for i, label in enumerate(labels):
             if label == 1:
                 total_diff = 0
@@ -490,6 +497,8 @@ def generate(  # pylint: disable=W0221
                                the start of perturbation region.
         :return x: our adversarial examples.
         """
+        import tensorflow as tf  # lgtm [py/repeated-import]
+
         # make copy so original data is not modified.
         adv_x = x.copy()
         if sample_sizes is None: