interpretable-classification

Introduces Fast Evidential Rule Learning (FERL), a method for interpretable classification that produces evidential outputs and can abstain when uncertain, with theoretical stability guarantees and strong empirical results across tabular and concept-bottleneck benchmarks.