Merge pull request #1762 from Trusted-AI/development_maintenance_111

Maintenance updates for ART 1.11.0

Author: beat-buesser

art/attacks/evasion/auto_attack.py

@@ -159,7 +159,7 @@ def generate(self, x: np.ndarray, y: Optional[np.ndarray] = None, **kwargs) -> n
         """
         x_adv = x.astype(ART_NUMPY_DTYPE)
         if y is not None:
-            y = check_and_transform_label_format(y, self.estimator.nb_classes)
+            y = check_and_transform_label_format(y, nb_classes=self.estimator.nb_classes)
 
         if y is None:
             y = get_labels_np_array(self.estimator.predict(x, batch_size=self.batch_size))
@@ -207,7 +207,9 @@ def generate(self, x: np.ndarray, y: Optional[np.ndarray] = None, **kwargs) -> n
                         if np.sum(sample_is_robust) == 0:
                             break
 
-                        target = check_and_transform_label_format(targeted_labels[:, i], self.estimator.nb_classes)
+                        target = check_and_transform_label_format(
+                            targeted_labels[:, i], nb_classes=self.estimator.nb_classes
+                        )
 
                         x_adv, sample_is_robust = self._run_attack(
                             x=x_adv,

art/attacks/evasion/auto_projected_gradient_descent.py

@@ -376,7 +376,7 @@ def generate(self, x: np.ndarray, y: Optional[np.ndarray] = None, **kwargs) -> n
         mask = kwargs.get("mask")
 
         if y is not None:
-            y = check_and_transform_label_format(y, self.estimator.nb_classes)
+            y = check_and_transform_label_format(y, nb_classes=self.estimator.nb_classes)
 
         if y is None:
             if self.targeted:

art/attacks/evasion/boundary.py

@@ -131,7 +131,7 @@ def generate(self, x: np.ndarray, y: Optional[np.ndarray] = None, **kwargs) -> n
             # Use model predictions as correct outputs
             y = get_labels_np_array(self.estimator.predict(x, batch_size=self.batch_size))  # type: ignore
 
-        y = check_and_transform_label_format(y, self.estimator.nb_classes, return_one_hot=False)
+        y = check_and_transform_label_format(y, nb_classes=self.estimator.nb_classes, return_one_hot=False)
 
         if y is not None and self.estimator.nb_classes == 2 and y.shape[1] == 1:
             raise ValueError(  # pragma: no cover

art/attacks/evasion/brendel_bethge.py

@@ -2186,7 +2186,7 @@ def generate(
         originals = x.copy()
 
         if y is not None:
-            y = check_and_transform_label_format(y, self.estimator.nb_classes)
+            y = check_and_transform_label_format(y, nb_classes=self.estimator.nb_classes)
 
         if y is None:
             # Throw error if attack is targeted, but no targets are provided

art/attacks/evasion/carlini.py

@@ -238,7 +238,7 @@ def generate(self, x: np.ndarray, y: Optional[np.ndarray] = None, **kwargs) -> n
         :return: An array holding the adversarial examples.
         """
         if y is not None:
-            y = check_and_transform_label_format(y, self.estimator.nb_classes)
+            y = check_and_transform_label_format(y, nb_classes=self.estimator.nb_classes)
         x_adv = x.astype(ART_NUMPY_DTYPE)
 
         if self.estimator.clip_values is not None:
@@ -743,7 +743,7 @@ def generate(self, x: np.ndarray, y: Optional[np.ndarray] = None, **kwargs) -> n
         :return: An array holding the adversarial examples.
         """
         if y is not None:
-            y = check_and_transform_label_format(y, self.estimator.nb_classes)
+            y = check_and_transform_label_format(y, nb_classes=self.estimator.nb_classes)
         x_adv = x.astype(ART_NUMPY_DTYPE)
 
         if self.estimator.clip_values is not None:
@@ -953,7 +953,7 @@ def generate(self, x: np.ndarray, y: Optional[np.ndarray] = None, **kwargs) -> n
         :return: An array holding the adversarial examples.
         """
         if y is not None:
-            y = check_and_transform_label_format(y, self.estimator.nb_classes)
+            y = check_and_transform_label_format(y, nb_classes=self.estimator.nb_classes)
         x_adv = x.astype(ART_NUMPY_DTYPE)
 
         if self.estimator.clip_values is not None:

art/attacks/evasion/decision_tree_attack.py

@@ -115,7 +115,7 @@ def generate(self, x: np.ndarray, y: Optional[np.ndarray] = None, **kwargs) -> n
         :return: An array holding the adversarial examples.
         """
         if y is not None:
-            y = check_and_transform_label_format(y, self.estimator.nb_classes, return_one_hot=False)
+            y = check_and_transform_label_format(y, nb_classes=self.estimator.nb_classes, return_one_hot=False)
         x_adv = x.copy()
 
         for index in trange(x_adv.shape[0], desc="Decision tree attack", disable=not self.verbose):

art/attacks/evasion/elastic_net.py

@@ -205,7 +205,7 @@ def generate(self, x: np.ndarray, y: Optional[np.ndarray] = None, **kwargs) -> n
         :return: An array holding the adversarial examples.
         """
         if y is not None:
-            y = check_and_transform_label_format(y, self.estimator.nb_classes)
+            y = check_and_transform_label_format(y, nb_classes=self.estimator.nb_classes)
         x_adv = x.astype(ART_NUMPY_DTYPE)
 
         # Assert that, if attack is targeted, y is provided:

art/attacks/evasion/fast_gradient.py

@@ -225,7 +225,7 @@ def generate(self, x: np.ndarray, y: Optional[np.ndarray] = None, **kwargs) -> n
 
         if isinstance(self.estimator, ClassifierMixin):
             if y is not None:
-                y = check_and_transform_label_format(y, self.estimator.nb_classes)
+                y = check_and_transform_label_format(y, nb_classes=self.estimator.nb_classes)
 
             if y is None:
                 # Throw error if attack is targeted, but no targets are provided

art/attacks/evasion/geometric_decision_based_attack.py

@@ -169,7 +169,7 @@ def generate(self, x: np.ndarray, y: Optional[np.ndarray] = None, **kwargs) -> n
         :return: The adversarial examples.
         """
         if y is not None:
-            y = check_and_transform_label_format(y, self.estimator.nb_classes, return_one_hot=True)
+            y = check_and_transform_label_format(y, nb_classes=self.estimator.nb_classes, return_one_hot=True)
 
         if y is not None and self.estimator.nb_classes == 2 and y.shape[1] == 1:  # pragma: no cover
             raise ValueError(

art/attacks/evasion/hop_skip_jump.py

@@ -133,7 +133,7 @@ def generate(self, x: np.ndarray, y: Optional[np.ndarray] = None, **kwargs) -> n
             # Use model predictions as correct outputs
             y = get_labels_np_array(self.estimator.predict(x, batch_size=self.batch_size))  # type: ignore
 
-        y = check_and_transform_label_format(y, self.estimator.nb_classes)
+        y = check_and_transform_label_format(y, nb_classes=self.estimator.nb_classes)
 
         if self.estimator.nb_classes == 2 and y.shape[1] == 1:  # pragma: no cover
             raise ValueError(