Attainability and Optimality: The Equalized Odds Fairness Revisited
It addresses the fundamental problem of ensuring fairness in ML predictions for stakeholders concerned with algorithmic bias, though it is incremental as it builds on existing fairness notions.
The paper investigates whether the Equalized Odds fairness criterion can be achieved in machine learning predictions, showing that it is attainable under certain conditions for deterministic functions and always for stochastic predictors with mild assumptions, and proves that training with all features yields better performance than post-processing while maintaining fairness.
Fairness of machine learning algorithms has been of increasing interest. In order to suppress or eliminate discrimination in prediction, various notions as well as approaches have been proposed to impose fairness. Given a notion of fairness, an essential problem is then whether or not it can always be attained, even if with an unlimited amount of data. This issue is, however, not well addressed yet. In this paper, focusing on the Equalized Odds notion of fairness, we consider the attainability of this criterion and, furthermore, if it is attainable, the optimality of the prediction performance under various settings. In particular, for prediction performed by a deterministic function of input features, we give conditions under which Equalized Odds can hold true; if the stochastic prediction is acceptable, we show that under mild assumptions, fair predictors can always be derived. For classification, we further prove that compared to enforcing fairness by post-processing, one can always benefit from exploiting all available features during training and get potentially better prediction performance while remaining fair. Moreover, while stochastic prediction can attain Equalized Odds with theoretical guarantees, we also discuss its limitation and potential negative social impacts.