Update KerasClassifier for verbose argument

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

Author: beat-buesser

art/estimators/classification/keras.py

@@ -559,7 +559,9 @@ def predict(  # pylint: disable=W0221
 
         return predictions
 
-    def fit(self, x: np.ndarray, y: np.ndarray, batch_size: int = 128, nb_epochs: int = 20, **kwargs) -> None:
+    def fit(
+        self, x: np.ndarray, y: np.ndarray, batch_size: int = 128, nb_epochs: int = 20, verbose: bool = False, **kwargs
+    ) -> None:
         """
         Fit the classifier on the training set `(x, y)`.
 
@@ -568,6 +570,7 @@ def fit(self, x: np.ndarray, y: np.ndarray, batch_size: int = 128, nb_epochs: in
                   shape (nb_samples,).
         :param batch_size: Size of batches.
         :param nb_epochs: Number of epochs to use for training.
+        :param verbose: Display training progress bar.
         :param kwargs: Dictionary of framework-specific arguments. These should be parameters supported by the
                `fit_generator` function in Keras and will be passed to this function as such. Including the number of
                epochs or the number of steps per epoch as part of this argument will result in as error.
@@ -582,18 +585,18 @@ def fit(self, x: np.ndarray, y: np.ndarray, batch_size: int = 128, nb_epochs: in
         if self._reduce_labels or y_ndim == 1:
             y_preprocessed = np.argmax(y_preprocessed, axis=1)
 
-        if "verbose" in kwargs:
-            kwargs["verbose"] = int(kwargs["verbose"])
-
-        self._model.fit(x=x_preprocessed, y=y_preprocessed, batch_size=batch_size, epochs=nb_epochs, **kwargs)
+        self._model.fit(
+            x=x_preprocessed, y=y_preprocessed, batch_size=batch_size, epochs=nb_epochs, verbose=int(verbose), **kwargs
+        )
 
-    def fit_generator(self, generator: "DataGenerator", nb_epochs: int = 20, **kwargs) -> None:
+    def fit_generator(self, generator: "DataGenerator", nb_epochs: int = 20, verbose: bool = False, **kwargs) -> None:
         """
         Fit the classifier using the generator that yields batches as specified.
 
         :param generator: Batch generator providing `(x, y)` for each epoch. If the generator can be used for native
                           training in Keras, it will.
         :param nb_epochs: Number of epochs to use for training.
+        :param verbose: Display training progress bar.
         :param kwargs: Dictionary of framework-specific arguments. These should be parameters supported by the
                `fit_generator` function in Keras and will be passed to this function as such. Including the number of
                epochs as part of this argument will result in as error.
@@ -603,9 +606,6 @@ def fit_generator(self, generator: "DataGenerator", nb_epochs: int = 20, **kwarg
         # Try to use the generator as a Keras native generator, otherwise use it through the `DataGenerator` interface
         from art.preprocessing.standardisation_mean_std.numpy import StandardisationMeanStd
 
-        if "verbose" in kwargs:
-            kwargs["verbose"] = int(kwargs["verbose"])
-
         if isinstance(generator, KerasDataGenerator) and (
             self.preprocessing is None
             or (
@@ -618,12 +618,12 @@ def fit_generator(self, generator: "DataGenerator", nb_epochs: int = 20, **kwarg
             )
         ):
             try:
-                self._model.fit_generator(generator.iterator, epochs=nb_epochs, **kwargs)
+                self._model.fit_generator(generator.iterator, epochs=nb_epochs, verbose=int(verbose), **kwargs)
             except ValueError:  # pragma: no cover
                 logger.info("Unable to use data generator as Keras generator. Now treating as framework-independent.")
-                super().fit_generator(generator, nb_epochs=nb_epochs, **kwargs)
+                super().fit_generator(generator, nb_epochs=nb_epochs, verbose=verbose, **kwargs)
         else:  # pragma: no cover
-            super().fit_generator(generator, nb_epochs=nb_epochs, **kwargs)
+            super().fit_generator(generator, nb_epochs=nb_epochs, verbose=verbose, **kwargs)
 
     def get_activations(
         self, x: np.ndarray, layer: Union[int, str], batch_size: int = 128, framework: bool = False

art/estimators/classification/pytorch.py

@@ -391,7 +391,7 @@ def fit(  # pylint: disable=W0221
                           the batch size. If ``False`` and the size of dataset is not divisible by the batch size, then
                           the last batch will be smaller. (default: ``False``)
         :param scheduler: Learning rate scheduler to run at the start of every epoch.
-        :param verbose: If to display the progress bar information.
+        :param verbose: Display training progress bar.
         :param kwargs: Dictionary of framework-specific arguments. This parameter is not currently supported for PyTorch
                        and providing it takes no effect.
         """

art/estimators/keras.py

@@ -61,8 +61,6 @@ def predict(self, x: np.ndarray, batch_size: int = 128, **kwargs):
         :return: Predictions.
         :rtype: Format as expected by the `model`
         """
-        if "verbose" in kwargs:
-            kwargs["verbose"] = int(kwargs["verbose"])
         return NeuralNetworkMixin.predict(self, x, batch_size=batch_size, **kwargs)
 
     def fit(self, x: np.ndarray, y, batch_size: int = 128, nb_epochs: int = 20, **kwargs) -> None:
@@ -76,8 +74,6 @@ def fit(self, x: np.ndarray, y, batch_size: int = 128, nb_epochs: int = 20, **kw
         :param batch_size: Batch size.
         :param nb_epochs: Number of training epochs.
         """
-        if "verbose" in kwargs:
-            kwargs["verbose"] = int(kwargs["verbose"])
         NeuralNetworkMixin.fit(self, x, y, batch_size=batch_size, nb_epochs=nb_epochs, **kwargs)
 
     def compute_loss(self, x: np.ndarray, y: np.ndarray, **kwargs) -> np.ndarray: