Adversarial Removal of Demographic Attributes from Text Data

Understand

Recent advances in Representation Learning and Adversarial Training seem to succeed in removing unwanted features from the learned representation.

• We show that demographic information of authors is encoded in -- and can be recovered from -- the intermediate representations learned by text-based neural classifiers.
• The implication is that decisions of classifiers trained on textual data are not agnostic to -- and likely condition on -- demographic attributes.
• When attempting to remove such demographic information using adversarial training, we find that while the adversarial component achieves chance-level development-set accuracy during training, a post-hoc classifier, trained on the encoded sentences from the first part, still manages to reach substantially higher classification accuracies on the same data.