Meta Optimality for Demographic Parity Constrained Regression via Post-Processing
This work addresses fairness constraints in machine learning for practitioners, offering a flexible approach to adapt existing regression techniques, though it is incremental as it builds on prior optimality analyses.
The paper tackles the problem of achieving demographic parity in regression by providing meta-theorems to validate fair minimax optimality across various data models and shows that optimal fair regression can be achieved through post-processing methods.
We address the regression problem under the constraint of demographic parity, a commonly used fairness definition. Recent studies have revealed fair minimax optimal regression algorithms, the most accurate algorithms that adhere to the fairness constraint. However, these analyses are tightly coupled with specific data generation models. In this paper, we provide meta-theorems that can be applied to various situations to validate the fair minimax optimality of the corresponding regression algorithms. Furthermore, we demonstrate that fair minimax optimal regression can be achieved through post-processing methods, allowing researchers and practitioners to focus on improving conventional regression techniques, which can then be efficiently adapted for fair regression.