Prediction-Powered Smoothing and Validation for Disaggregated AI Evaluation
A Bayesian model fit to each domain's prediction-powered estimate, with an extension that borrows strength across a reporting taxonomy, and a new, approximately unbiased design-based cross-validation score for choosing among direct and smoothed estimators for estimation and validation are derived.
Sho Kawano, Z. Li, Paul A. Parker
· 0 citations