Reproduce “The Sensitivity of Counterfactual Fairness to Unmeasured Confounding” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Niki Kilbertus, Philip J. Ball, Matt J. Kusner, Adrian Weller, Ricardo Silva
The authors introduce a sensitivity analysis for counterfactually fair predictors in additive noise models, representing omitted confounding through bounded correlations between errors. It measures counterfactual unfairness (CFU) as the average squared difference between factual and counterfactual predictions generated under those correlations. In two datasets, measures stayed below reference predictors’ observed-data deviations; this does not establish real-world fairness or true confounding.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.