b09284f6ba59659a9819e68244a7a785016c87c5,art/attacks/evasion/shadow_attack.py,ShadowAttack,generate,#ShadowAttack#Any#Any#,119

Before Change

        x_adv = np.zeros_like(x, dtype=ART_NUMPY_DTYPE)

        // Compute perturbation with implicit batching
        for i_batch in range(int(np.ceil(x.shape[0] / self.batch_size))):

            batch_index_1, batch_index_2 = i_batch * self.batch_size, (i_batch + 1) * self.batch_size
            x_batch = x[batch_index_1:batch_index_2]
            y_batch = y[batch_index_1:batch_index_2]

            perturbation = (
                np.random.uniform(
                    low=self.estimator.clip_values[0], high=self.estimator.clip_values[1], size=x_batch.shape
                ).astype(ART_NUMPY_DTYPE)
                - (self.estimator.clip_values[1] - self.estimator.clip_values[0]) / 2
            )

            for _ in range(self.nb_steps):

                gradients_ce = self.estimator.loss_gradient(x=x_batch + perturbation, y=y_batch, sampling=False) * (
                    1 - 2 * int(self.targeted)
                )

                gradients = gradients_ce - self._get_regularisation_loss_gradients(perturbation)

                perturbation += self.learning_rate * gradients

            x_p = x_batch + perturbation

            x_p = np.clip(x_p, a_min=self.estimator.clip_values[0], a_max=self.estimator.clip_values[1])

            perturbation = x_p - x_batch

            x_adv[batch_index_1:batch_index_2] = x_batch + perturbation

        return x_adv

    def _get_regularisation_loss_gradients(self, perturbation):

After Change

        x = x.astype(ART_NUMPY_DTYPE)

        x_batch = np.repeat(x, repeats=self.batch_size, axis=0)

        x_batch = x_batch + np.random.normal(scale=self.sigma, size=x_batch)

        y_batch = np.repeat(y, repeats=self.batch_size, axis=0)

        perturbation = (

Instances