A Human-Centered Validation of the Explainability-Performance Coefficient
2026-07-31 • Machine Learning
Machine LearningArtificial IntelligenceComputer Vision and Pattern Recognition
AI summaryⓘ
The authors address the challenge of measuring how good explanations from AI models are, especially when these models are used in important areas. They created a new way to score explanations called the EPC score, which balances using fewer features with keeping the model accurate. They tested this score on different kinds of data, like tables, text, and images, and found it helps understand how explanations work. They also showed that the EPC score matches well with how humans explain things in words or pictures.
Explainable Artificial IntelligenceXAIExplanation FidelityEPC ScoreFeature SelectionModel PerformanceTabular DataText DataImage DataHuman Evaluation
Authors
Christian Oliva, Luis F. Lago-Fernández
Abstract
The rapid adoption of deep learning models in high-risk domains has intensified the need for trustworthy Explainable Artificial Intelligence (XAI). However, objectively evaluating explanation fidelity and aligning XAI metrics with human-centered understanding remain critical open challenges. In this work, we propose a model-agnostic metric, the EPC score, which is an extension of the Explainability-Performance Coefficient (EPC), that quantifies explanation quality by explicitly balancing the trade-off between feature selection sparsity and preserved model performance. Through an empirical validation across tabular, text, and image modalities, we show that the EPC score effectively uncovers operational dependencies among network activations, data dimensionality, and explainer performance. Furthermore, we validate the EPC score against independent human-based explanations, proving that higher EPC scores strongly align with human lexical sentiment judgments and spatial visual annotations.