ML technique aimed at reducing bias in models by using adversarial training, where one network tries to predict sensitive attributes and another tries to prevent it.
Generality: 659
