AI Paper Atlas

Research map / Graph learning / GNN architectures

Higher-Order Explanations of Graph Neural Networks via Relevant Walks

Thomas Schnake, Oliver Eberle, Jonas Lederer, Shinichi Nakajima, Kristof T. Schütt, Klaus‐Robert Müller, Grégoire Montavon

TPAMI 2021 · Graph learning › GNN architectures · approach: GNN explainability · 325 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use LRP-based relevant-walk propagation instead of gradient-times-input (GNN-LRP vs GNN-GI) (ablation)higher-order GNN explanation · Synthetic Spectral, activation taskAUFC: 0.87 -> 2.29Improved
use GNN-LRP instead of GNN-GI (ablation)higher-order GNN explanation · VGG-16, pruning taskAUFC: 7.78 -> 6.05Improved
use GNN-LRP instead of GNN-GI (ablation)higher-order GNN explanation · VGG-16, activation taskAUFC: 12.03 -> 14.04Improved
use relevant-walk explanation instead of first-order Pope LRPGNN explanation · Synthetic Spectral, activation taskAUFC: 1.39 -> 2.29Improved
use GNN-LRP instead of GNNExplainerGNN explanation · Synthetic GIN, pruning taskAUFC: 3.05 -> 2.85Improved
use GNN-LRP instead of GNNExplainerGNN explanation · Sentiment GCN, activation taskAUFC: 20.23 -> 20.05No clear effect
use GNN-LRP instead of GNNExplainerGNN explanation · SchNet-E, activation taskAUFC: 10.41 -> 10.47No clear effect
use GNN-LRP instead of GNNExplainerGNN explanation · SchNet-mu, pruning taskAUFC: 0.28 -> 0.26Improved
use GNN-LRP instead of GNNExplainerGNN explanation · SchNet-mu, activation taskAUFC: 1.01 -> 1.09Improved
use GNN-LRP instead of random explanationGNN explanation · Synth-GIN, activation taskAUFC: 1.39 -> 3.92Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy