Journal
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
Volume 45, Issue 1, Pages 525-538Publisher
IEEE COMPUTER SOC
DOI: 10.1109/TPAMI.2022.3148905
Keywords
Fair representation; group fairness; fair machine learning
Ask authors/readers for more resources
This study investigates whether fair representation can ensure fairness for unknown tasks and multiple fairness notions. It is found that fair representation guarantees fairness for an important subset of tasks where the representation is discriminative. Experimental results show that the learned representation indeed leads to fairer downstream predictions that are unknown during the representation learning stage.
Motivated by scenarios where data is used for diverse prediction tasks, we study whether fair representation can be used to guarantee fairness for unknown tasks and for multiple fairness notions. We consider seven group fairness notions that cover the concepts of independence, separation, and calibration. Against the backdrop of the fairness impossibility results, we explore approximate fairness. We prove that, although fair representation might not guarantee fairness for all prediction tasks, it does guarantee fairness for an important subset of tasks-the tasks for which the representation is discriminative. Specifically, all seven group fairness notions are linearly controlled by fairness and discriminativeness of the representation. When an incompatibility exists between different fairness notions, fair and discriminative representation hits the sweet spot that approximately satisfies all notions. Motivated by our theoretical findings, we propose to learn both fair and discriminative representations using pretext loss which self-supervises learning, and Maximum Mean Discrepancy as a fair regularizer. Experiments on tabular, image, and face datasets show that using the learned representation, downstream predictions that we are unaware of when learning the representation indeed become fairer. The fairness guarantees computed from our theoretical results are all valid.
Authors
I am an author on this paper
Click your name to claim this paper and add it to your profile.
Reviews
Recommended
No Data Available