Preserving Fairness in AI under Domain Shift

Serban Stan, Mohammad Rostami

Existing algorithms for ensuring fairness in AI use a single-shot training strategy, where an AI model is trained on an annotated training dataset with sensitive attributes and the…