Explainability of Complex AI Models with Correlation Impact Ratio

2026-01-10

arXiv.org

AI Summary

This work addresses the limitations of existing post-hoc explanation methods, which often produce unreliable feature rankings when features are highly correlated and rely on costly perturbations that hinder scalability to high-dimensional data. To overcome these issues, the authors propose ExCIR (Explainability via Correlation Influence Ratio), a theoretically grounded and computationally efficient approach that evaluates feature contributions in a single forward pass without perturbations while explicitly modeling inter-feature correlations. Built within an information-theoretic framework, ExCIR unifies the correlation ratio and canonical correlation analysis, enabling explanations for multi-output and class-conditional settings. Experimental results on diverse datasets—including EEG, synthetic vehicle, Digits, and Cats-Dogs—demonstrate that ExCIR consistently outperforms mainstream methods such as LIME and SHAP in terms of stability, interpretability, and computational efficiency.

Technology Category

• Machine Learning: Feature Construction/Reformulation
• Computer Vision: Interpretability, Explainability, and Transparency
• Humans and AI: Explainable AI (XAI) for Human Understanding

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Complex AI systems make better predictions but often lack transparency, limiting trustworthiness, interpretability, and safe deployment. Common post hoc AI explainers, such as LIME, SHAP, HSIC, and SAGE, are model agnostic but are too restricted in one significant regard: they tend to misrank correlated features and require costly perturbations, which do not scale to high dimensional data. We introduce ExCIR (Explainability through Correlation Impact Ratio), a theoretically grounded, simple, and reliable metric for explaining the contribution of input features to model outputs, which remains stable and consistent under noise and sampling variations. We demonstrate that ExCIR captures dependencies arising from correlated features through a lightweight single pass formulation. Experimental evaluations on diverse datasets, including EEG, synthetic vehicular data, Digits, and Cats-Dogs, validate the effectiveness and stability of ExCIR across domains, achieving more interpretable feature explanations than existing methods while remaining computationally efficient. To this end, we further extend ExCIR with an information theoretic foundation that unifies the correlation ratio with Canonical Correlation Analysis under mutual information bounds, enabling multi output and class conditioned explainability at scale.

Problem

• Explainability
• Correlated Features
• Post hoc Explanation
• High-dimensional Data
• Feature Attribution

Innovation

• ExCIR
• Correlation Impact Ratio
• Model Explainability
• Feature Attribution
• Information-Theoretic Explanation