ScoreCAM GNN: une explication optimale des r\'eseaux profonds sur graphes

TL;DR

This paper introduces ScoreCAM GNN, an improved method for explaining deep graph neural networks that is more optimal, lightweight, consistent, and better utilizes graph topology than existing approaches.

Contribution

The paper presents ScoreCAM GNN, a novel explanation method specifically designed for graph neural networks, enhancing interpretability and performance over prior techniques.

Findings

More optimal explanation quality

Lighter computational complexity

Better exploitation of graph topology

Abstract

The explainability of deep networks is becoming a central issue in the deep learning community. It is the same for learning on graphs, a data structure present in many real world problems. In this paper, we propose a method that is more optimal, lighter, consistent and better exploits the topology of the evaluated graph than the state-of-the-art methods.