Intersectional Divergence: Measuring Fairness in Regression
This addresses fairness gaps in regression for ML practitioners, though it's incremental as it extends existing fairness concepts to regression.
The paper tackles the problem of measuring fairness in regression tasks by proposing Intersectional Divergence (ID), which considers combinations of all protected attributes and differentiates prediction impacts in user-relevant target ranges, showing that incorporating IDLoss into optimization can considerably improve fairness while maintaining competitive predictive performance.
Fairness in machine learning research is commonly framed in the context of classification tasks, leaving critical gaps in regression. In this paper, we propose a novel approach to measure intersectional fairness in regression tasks, going beyond the focus on single protected attributes from existing work to consider combinations of all protected attributes. Furthermore, we contend that it is insufficient to measure the average error of groups without regard for imbalanced domain preferences. Accordingly, we propose Intersectional Divergence (ID) as the first fairness measure for regression tasks that 1) describes fair model behavior across multiple protected attributes and 2) differentiates the impact of predictions in target ranges most relevant to users. We extend our proposal demonstrating how ID can be adapted into a loss function, IDLoss, that satisfies convergence guarantees and has piecewise smooth properties that enable practical optimization. Through an extensive experimental evaluation, we demonstrate how ID allows unique insights into model behavior and fairness, and how incorporating IDLoss into optimization can considerably improve single-attribute and intersectional model fairness while maintaining a competitive balance in predictive performance.