Imbalanced Classification through the Lens of Spurious Correlations

TL;DR

This paper introduces a novel approach using Explainable AI to identify and eliminate spurious correlations caused by class imbalance, improving classification reliability and providing new insights into imbalance effects.

Contribution

It presents a counterfactual explanations-based method to detect and mitigate Clever Hans effects in imbalanced datasets, a perspective not addressed by previous techniques.

Findings

Achieves competitive classification performance on three datasets.

Demonstrates emergence of Clever Hans effects under class imbalance.

Provides a new perspective on imbalance-related spurious correlations.

Abstract

Class imbalance poses a fundamental challenge in machine learning, frequently leading to unreliable classification performance. While prior methods focus on data- or loss-reweighting schemes, we view imbalance as a data condition that amplifies Clever Hans (CH) effects by underspecification of minority classes. In a counterfactual explanations-based approach, we propose to leverage Explainable AI to jointly identify and eliminate CH effects emerging under imbalance. Our method achieves competitive classification performance on three datasets and demonstrates how CH effects emerge under imbalance, a perspective largely overlooked by existing approaches.