Long Term Fairness for Minority Groups via Performative Distributionally Robust Optimization
This work tackles fairness issues in ML for minority groups, but appears incremental as it builds on existing performative prediction methods.
The paper addresses limitations in existing fairness criteria for machine learning models by extending performative prediction with a distributionally robust objective to improve long-term fairness for minority groups.
Fairness researchers in machine learning (ML) have coalesced around several fairness criteria which provide formal definitions of what it means for an ML model to be fair. However, these criteria have some serious limitations. We identify four key shortcomings of these formal fairness criteria, and aim to help to address them by extending performative prediction to include a distributionally robust objective.