Reproduce “Pragmatic Fairness: Developing Policies with Outcome Disparity Control” 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
Limor Gultchin, Siyuan Guo, Alan Malek, Silvia Chiappa, Ricardo Silva
The authors introduce a causal framework for optimizing actions to improve downstream outcomes using historical data and a bounded action space. It offers two fairness targets: reducing action-moderated group differences, or equalizing policy gains relative to a historical baseline. In semi-synthetic experiments, the targets produce different utility–constraint patterns; these modeled results do not establish effects in real-world deployments.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.