Jul 2026
RUBRIC: Realism-Utility Balanced Ranking for Imbalanced Classification
This work introduces RUBRIC, a generator-agnostic filtering framework that formulates synthetic sample selection as a quality-over-quantity optimization problem and shows that RUBRIC improves F1-macro and recall while maintaining comparable ROC-AUC across several generators.
Yan-Xuan Yu, Dong Liu, Shu Wang et al.
· arXiv.org · 1 citation