COBRA: Counterfactual oversampling framework for imbalanced structured data classification

doi:10.1016/j.eswa.2025.130764

COBRA: Counterfactual oversampling framework for imbalanced structured data classification

2025-12-15

0

PRE

AI

DOI：10.1016/j.eswa.2025.130764

Original

Original request for help

Save

Abstract

En 中文

highlights • Counterfactual oversampling (COBRA-IDS) to solve the problem of rare attacks in cybersecurity datasets. • Boundary-focused, plausible counterfactuals to improve minority-class detection vs. traditional techniques. • XAI analysis. • Robustness of the proposed method to adversarial attacks.

COBRA: Counterfactual oversampling framework for imbalanced structured data classification

Abstract

Journal

Organization