Causal Fair Machine Learning via Rank-Preserving Interventional Distributions
This addresses fairness in machine learning for automated decision systems, but it is incremental as it builds on prior causal approaches.
The paper tackles unfairness in automated decision-making by proposing a causal fairness method using rank-preserving interventional distributions to define a normatively desired world where protected attributes have no effect, and shows through experiments that it effectively identifies discriminated individuals and mitigates unfairness compared to an existing method.
A decision can be defined as fair if equal individuals are treated equally and unequals unequally. Adopting this definition, the task of designing machine learning (ML) models that mitigate unfairness in automated decision-making systems must include causal thinking when introducing protected attributes: Following a recent proposal, we define individuals as being normatively equal if they are equal in a fictitious, normatively desired (FiND) world, where the protected attributes have no (direct or indirect) causal effect on the target. We propose rank-preserving interventional distributions to define a specific FiND world in which this holds and a warping method for estimation. Evaluation criteria for both the method and the resulting ML model are presented and validated through simulations. Experiments on empirical data showcase the practical application of our method and compare results with "fairadapt" (Plečko and Meinshausen, 2020), a different approach for mitigating unfairness by causally preprocessing data that uses quantile regression forests. With this, we show that our warping approach effectively identifies the most discriminated individuals and mitigates unfairness.