Learning Adversarially Fair and Transferable Representations
This work addresses fairness in machine learning for scenarios where third parties use learned representations with unknown objectives, offering a method to ensure fair predictions across tasks.
The paper tackled the problem of unfair prediction outcomes in downstream tasks by using adversarial representation learning to enforce group fairness, showing that the choice of adversarial objective is crucial and demonstrating fair transfer learning with maintained utility.
In this paper, we advocate for representation learning as the key to mitigating unfair prediction outcomes downstream. Motivated by a scenario where learned representations are used by third parties with unknown objectives, we propose and explore adversarial representation learning as a natural method of ensuring those parties act fairly. We connect group fairness (demographic parity, equalized odds, and equal opportunity) to different adversarial objectives. Through worst-case theoretical guarantees and experimental validation, we show that the choice of this objective is crucial to fair prediction. Furthermore, we present the first in-depth experimental demonstration of fair transfer learning and demonstrate empirically that our learned representations admit fair predictions on new tasks while maintaining utility, an essential goal of fair representation learning.