def _test_backend_mnist(self, classifier, x_test, y_test): x_test_original = x_test.copy() df = VirtualAdversarialMethod(classifier, batch_size=100) from art.classifiers import TensorFlowClassifier if isinstance(classifier, TensorFlowClassifier): with self.assertRaises(TypeError) as context: x_test_adv = df.generate(x_test) self.assertIn('This attack requires a classifier predicting probabilities in the range [0, 1] as output.' 'Values smaller than 0.0 or larger than 1.0 have been detected.', str(context.exception)) else: x_test_adv = df.generate(x_test) self.assertFalse((x_test == x_test_adv).all()) y_pred = get_labels_np_array(classifier.predict(x_test_adv)) self.assertFalse((y_test == y_pred).all()) acc = np.sum(np.argmax(y_pred, axis=1) == np.argmax(y_test, axis=1)) / y_test.shape[0] logger.info('Accuracy on adversarial examples: %.2f%%', (acc * 100)) # Check that x_test has not been modified by attack and classifier self.assertAlmostEqual(float(np.max(np.abs(x_test_original - x_test))), 0.0, delta=0.00001)
def _test_backend_mnist(self, classifier): # Get MNIST (_, _), (x_test, y_test) = self.mnist x_test, y_test = x_test[:NB_TEST], y_test[:NB_TEST] df = VirtualAdversarialMethod(classifier, batch_size=100) from art.classifiers import TensorFlowClassifier if isinstance(classifier, TensorFlowClassifier): with self.assertRaises(TypeError) as context: x_test_adv = df.generate(x_test) self.assertIn( 'This attack requires a classifier predicting probabilities in the range [0, 1] as output.' 'Values smaller than 0.0 or larger than 1.0 have been detected.', str(context.exception)) else: x_test_adv = df.generate(x_test) self.assertFalse((x_test == x_test_adv).all()) y_pred = get_labels_np_array(classifier.predict(x_test_adv)) self.assertFalse((y_test == y_pred).all()) acc = np.sum( np.argmax(y_pred, axis=1) == np.argmax( y_test, axis=1)) / y_test.shape[0] logging.info('Accuracy on adversarial examples: %.2f%%', (acc * 100))
def test_pytorch_iris(self): (_, _), (x_test, y_test) = self.iris classifier = get_iris_classifier_pt() attack = VirtualAdversarialMethod(classifier, eps=.1) with self.assertRaises(TypeError) as context: x_test_adv = attack.generate(x_test.astype(np.float32)) self.assertIn('This attack requires a classifier predicting probabilities in the range [0, 1] as output.' 'Values smaller than 0.0 or larger than 1.0 have been detected.', str(context.exception))
def test_tensorflow_iris(self): classifier, _ = get_tabular_classifier_tf() attack = VirtualAdversarialMethod(classifier, eps=0.1) with self.assertRaises(TypeError) as context: x_test_iris_adv = attack.generate(self.x_test_iris) self.assertIn( "This attack requires a classifier predicting probabilities in the range [0, 1] as output." "Values smaller than 0.0 or larger than 1.0 have been detected.", str(context.exception), )
def test_keras_iris_clipped(self): classifier = get_tabular_classifier_kr() # Test untargeted attack attack = VirtualAdversarialMethod(classifier, eps=0.1) x_test_iris_adv = attack.generate(self.x_test_iris) self.assertFalse((self.x_test_iris == x_test_iris_adv).all()) self.assertTrue((x_test_iris_adv <= 1).all()) self.assertTrue((x_test_iris_adv >= 0).all()) preds_adv = np.argmax(classifier.predict(x_test_iris_adv), axis=1) self.assertFalse((np.argmax(self.y_test_iris, axis=1) == preds_adv).all()) acc = np.sum(preds_adv == np.argmax(self.y_test_iris, axis=1)) / self.y_test_iris.shape[0] logger.info("Accuracy on Iris with VAT adversarial examples: %.2f%%", (acc * 100))
def test_keras_iris_unbounded(self): classifier = get_tabular_classifier_kr() # Recreate a classifier without clip values classifier = KerasClassifier(model=classifier._model, use_logits=False, channel_index=1) attack = VirtualAdversarialMethod(classifier, eps=1) x_test_iris_adv = attack.generate(self.x_test_iris) self.assertFalse((self.x_test_iris == x_test_iris_adv).all()) self.assertTrue((x_test_iris_adv > 1).any()) self.assertTrue((x_test_iris_adv < 0).any()) preds_adv = np.argmax(classifier.predict(x_test_iris_adv), axis=1) self.assertFalse((np.argmax(self.y_test_iris, axis=1) == preds_adv).all()) acc = np.sum(preds_adv == np.argmax(self.y_test_iris, axis=1)) / self.y_test_iris.shape[0] logger.info("Accuracy on Iris with VAT adversarial examples: %.2f%%", (acc * 100))
def test_keras_iris_clipped(self): (_, _), (x_test, y_test) = self.iris classifier = get_iris_classifier_kr() # Test untargeted attack attack = VirtualAdversarialMethod(classifier, eps=.1) x_test_adv = attack.generate(x_test) self.assertFalse((x_test == x_test_adv).all()) self.assertTrue((x_test_adv <= 1).all()) self.assertTrue((x_test_adv >= 0).all()) preds_adv = np.argmax(classifier.predict(x_test_adv), axis=1) self.assertFalse((np.argmax(y_test, axis=1) == preds_adv).all()) acc = np.sum(preds_adv == np.argmax(y_test, axis=1)) / y_test.shape[0] logger.info('Accuracy on Iris with VAT adversarial examples: %.2f%%', (acc * 100))
def _test_backend_mnist(self, classifier, x_test, y_test): x_test_original = x_test.copy() df = VirtualAdversarialMethod(classifier, batch_size=100, max_iter=2) x_test_adv = df.generate(x_test) self.assertFalse((x_test == x_test_adv).all()) y_pred = get_labels_np_array(classifier.predict(x_test_adv)) self.assertFalse((y_test == y_pred).all()) acc = np.sum(np.argmax(y_pred, axis=1) == np.argmax(y_test, axis=1)) / y_test.shape[0] logger.info("Accuracy on adversarial examples: %.2f%%", (acc * 100)) # Check that x_test has not been modified by attack and classifier self.assertAlmostEqual(float(np.max(np.abs(x_test_original - x_test))), 0.0, delta=0.00001)