CAMs as Shapley Value-based Explainers

2025-01-09

AI Summary

Class Activation Mapping (CAM) methods lack clear theoretical foundations and rigorous justification. Method: This paper proposes the Content-Retentive Game-theoretic (CRG) framework, modeling CNN predictions as cooperative games and establishing, for the first time, a rigorous theoretical connection between CAM and Shapley values. Building upon CRG, we introduce ShapleyCAM: it employs a Residual Softmax Target-Class (ReST) utility function to correct softmax-induced bias and derives a closed-form approximation of Shapley values via gradient- and Hessian-driven second-order Taylor expansion—balancing theoretical soundness and computational efficiency. Contribution/Results: Evaluated on 12 mainstream models over ImageNet, ShapleyCAM significantly improves attribution localization accuracy and interpretability. The implementation is publicly available, bridging the gap between heuristic CAM approaches and principled, Shapley-based explanation methods.

Technology Category

• Computer Vision: Interpretability, Explainability, and Transparency
• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Causality

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Class Activation Mapping (CAM) methods are widely used to visualize neural network decisions, yet their underlying mechanisms remain incompletely understood. To enhance the understanding of CAM methods and improve their explainability, we introduce the Content Reserved Game-theoretic (CRG) Explainer. This theoretical framework clarifies the theoretical foundations of GradCAM and HiResCAM by modeling the neural network prediction process as a cooperative game. Within this framework, we develop ShapleyCAM, a new method that leverages gradients and the Hessian matrix to provide more precise and theoretically grounded visual explanations. Due to the computational infeasibility of exact Shapley value calculation, ShapleyCAM employs a second-order Taylor expansion of the cooperative game's utility function to derive a closed-form expression. Additionally, we propose the Residual Softmax Target-Class (ReST) utility function to address the limitations of pre-softmax and post-softmax scores. Extensive experiments across 12 popular networks on the ImageNet validation set demonstrate the effectiveness of ShapleyCAM and its variants. Our findings not only advance CAM explainability but also bridge the gap between heuristic-driven CAM methods and compute-intensive Shapley value-based methods. The code is available at url{https://github.com/caihuaiguang/pytorch-shapley-cam}.

Problem

• Class Activation Mapping
• Neural Network Interpretability
• Explainable AI

Innovation

• Content Preservation Game Theory (CRG)
• ShapleyCAM
• ReST Tool