Owen-based Semantics and Hierarchy-Aware Explanation (O-Shap)

2026-02-19

AI Summary

This work addresses a critical limitation of existing Shapley value–based visual explanation methods, which often fail due to their neglect of spatial and semantic dependencies among pixels and the use of grouping strategies that undermine consistency. To overcome this, the authors propose O-Shap, a novel approach that integrates Owen values within the SHAP framework through a hierarchical feature grouping mechanism satisfying the T-consistency property. This design ensures semantic alignment across attribution levels while enabling efficient pruning. As a result, O-Shap significantly improves attribution accuracy, semantic coherence, and computational efficiency. Extensive experiments demonstrate its consistent superiority over current SHAP variants on both image and tabular data.

Technology Category

• Computer Vision: Interpretability, Explainability, and Transparency
• Humans and AI: Explainable AI (XAI) for Human Understanding
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Security and Privacy: Data transparency and provenance
• 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

Abstract

Shapley value-based methods have become foundational in explainable artificial intelligence (XAI), offering theoretically grounded feature attributions through cooperative game theory. However, in practice, particularly in vision tasks, the assumption of feature independence breaks down, as features (i.e., pixels) often exhibit strong spatial and semantic dependencies. To address this, modern SHAP implementations now include the Owen value, a hierarchical generalization of the Shapley value that supports group attributions. While the Owen value preserves the foundations of Shapley values, its effectiveness critically depends on how feature groups are defined. We show that commonly used segmentations (e.g., axis-aligned or SLIC) violate key consistency properties, and propose a new segmentation approach that satisfies the $T$-property to ensure semantic alignment across hierarchy levels. This hierarchy enables computational pruning while improving attribution accuracy and interpretability. Experiments on image and tabular datasets demonstrate that O-Shap outperforms baseline SHAP variants in attribution precision, semantic coherence, and runtime efficiency, especially when structure matters.

Problem

• Shapley value
• feature dependence
• Owen value
• feature grouping
• semantic coherence

Innovation

• Owen value
• hierarchical explanation
• T-property
• feature grouping
• SHAP