Reproduce “Fair Regression with Wasserstein Barycenters” 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
Evgenii Chzhen, Christophe Denis, Mohamed Hebiri, Luca Oneto, Massimiliano Pontil
The authors characterize the squared-error-optimal Demographic-Parity predictor as a one-dimensional Wasserstein barycenter of groups’ baseline prediction distributions. Under a density condition, their post-processing method matches group-relative ranks and averages predictions by group prevalence. On NLSY GPA prediction, RLS+Ours shifts test MSE from 0.156 to 0.203 and KS from 0.50 to 0.09; this is one comparison, not a universal result.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.