e3a249cd2de8b2518470021db0f579e26cafbfba,aif360/sklearn/inprocessing/adversarial_debiasing.py,AdversarialDebiasing,decision_function,#AdversarialDebiasing#,185

Before Change

        n_samples = X.shape[0]
        groups, _ = check_groups(X, self.prot_attr_)
        le = LabelEncoder().fit(self.groups_)
        groups = le.transform(groups)

        samples_covered = 0
        scores = np.empty((n_samples, len(self.classes_)))
        while samples_covered < n_samples:
            start = samples_covered
            end = samples_covered + self.batch_size
            if end > n_samples:
                end = n_samples

            batch_ids = np.arange(start, end)
            batch_features = X[batch_ids]
            batch_prot_attr = groups[batch_ids]

            batch_feed_dict = {self.input_ph: batch_features,
                               self.keep_prob: 1.0}

After Change

                                              feed_dict=batch_feed_dict)
            samples_covered += len(batch_features)

        return scores.ravel() if scores.shape[1] == 1 else scores

    def predict_proba(self, X):
        decision = self.decision_function(X)

Instances