RUBRIC: Realism--Utility Balanced Ranking for Imbalanced Classification

2026-07-10

AI Summary

This work addresses the degradation of generalization performance and decision boundary distortion caused by low-quality synthetic samples generated through oversampling under class imbalance. To this end, we propose a generator-agnostic filtering framework that formulates synthetic sample selection as a quality-driven optimization problem. Our approach innovatively integrates discriminator-based authenticity with utility derived from a concave margin function to construct a ranking criterion that monotonically tightens the generalization bound of margin classifiers. A λ-weighted mechanism enables controllable filtering of synthetic instances. Experimental results demonstrate that the proposed framework significantly improves F1-macro and recall on imbalanced datasets such as credit card fraud detection, while maintaining stable ROC-AUC performance; moreover, adjusting λ effectively restores AUPRC.

Abstract

Class imbalance poses a fundamental challenge in risk-sensitive applications such as fraud detection and medical diagnosis, where minority-class samples are scarce yet critical for accurate classification. Existing oversampling methods generate synthetic samples to rebalance class distributions; however, they often produce large numbers of low-quality candidates that distort decision boundaries or introduce artifacts, leading to overfitting and degraded generalization. In this work, we introduce RUBRIC, a generator-agnostic filtering framework that formulates synthetic sample selection as a quality-over-quantity optimization problem. RUBRIC ranks candidates using a realism-utility trade-off: realism is quantified by a learned discriminator that distinguishes real samples from synthetic samples, while utility captures proximity to the decision boundary through a concave margin-based scoring function. We show that, under mild regularity conditions, the proposed filtering strategy monotonically tightens the generalization bound for margin-based classifiers by jointly reducing distribution shift and suppressing near-negative tail contributions. Through extensive experiments on credit-card fraud detection and other imbalanced benchmarks, we demonstrate that RUBRIC improves F1-macro and recall while maintaining comparable ROC-AUC across several generators. We also provide explicit lambda-sensitivity analysis to show how users can recover AUPRC when ranking quality is prioritized.

Problem

• class imbalance
• oversampling
• synthetic samples
• decision boundary distortion
• generalization degradation

Innovation

• class imbalance
• synthetic sample filtering
• realism-utility trade-off
• margin-based classification
• generalization bound