MI CAM: Mutual Information Weighted Activation Mapping for Causal Visual Explanations of Convolutional Neural Networks

TL;DR

MI CAM is a novel visual explanation method for CNNs that uses mutual information to produce causal, unbiased saliency maps, outperforming existing techniques in qualitative and quantitative evaluations.

Contribution

The paper introduces MI CAM, a new post-hoc explanation technique leveraging mutual information for causal and unbiased visualizations of CNN decisions.

Findings

MI CAM produces saliency maps comparable or superior to state-of-the-art methods.

It provides causal interpretations validated by counterfactual analysis.

Outperforms existing methods in qualitative and quantitative assessments.

Abstract

With the intervention of machine vision in our crucial day to day necessities including healthcare and automated power plants, attention has been drawn to the internal mechanisms of convolutional neural networks, and the reason why the network provides specific inferences. This paper proposes a novel post-hoc visual explanation method called MI CAM based on activation mapping. Differing from previous class activation mapping based approaches, MI CAM produces saliency visualizations by weighing each feature map through its mutual information with the input image and the final result is generated by a linear combination of weights and activation maps. It also adheres to producing causal interpretations as validated with the help of counterfactual analysis. We aim to exhibit the visual performance and unbiased justifications for the model inferencing procedure achieved by MI CAM. Our approach works at par with all state-of-the-art methods but particularly outperforms some in terms of qualitative and quantitative measures. The implementation of proposed method can be found on https://anonymous.4open.science/r/MI-CAM-4D27