CAMs as Shapley Value-based Explainers

TL;DR

This paper introduces a game-theoretic framework called CRG Explainer to clarify and improve CAM methods, developing ShapleyCAM which uses gradients and Hessians for more precise explanations, validated on ImageNet.

Contribution

It provides a theoretical foundation for CAM methods using Shapley values and introduces a new method, ShapleyCAM, with a closed-form solution for better explanations.

Findings

ShapleyCAM offers more precise visual explanations.

The framework bridges heuristic CAM methods and Shapley value-based approaches.

Extensive experiments validate the effectiveness of the proposed methods.

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}.