Fairness in Causal Bayesian Networks
A Causal Bayesian Networks Viewpoint on Fairness
We show that causal bayesian networks provide us with a powerful tool to measure unfairness in a dataset and to design fair models in complex unfairnessscenarios.We offer a graphical interpretation of unfairness in a dataset as the presence of an unfair causal path in the causal bayesian network representing the data generation mechanism.We use this viewpoint to revisit the recentdebate surrounding the compas pretrial risk assessment tool and, more generally, to point out that fairness evaluation on a model requires carefulconsiderations on the patterns of unfairness underlying the training data.
