AI Paper Atlas

Research map / Graph learning / GNN architectures

Paradigm Shift of GNN Explainer from Label Space to Prototypical Representation Space

Jun Yin, Senzhang Wang, Ziluowen Luo, Peng Huo, Hao Yan, Hao Miao, Chaozhuo Li, Shirui Pan, Chengqi Zhang

ICLR 2026 · Graph learning › GNN architectures · approach: GNN explainability

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
shift explainer optimization from label space to prototypical representation space (IDEA)PGExplainer · Average over 5 datasetsROC-AUC: 0.8000 -> 0.8856Improved
add IDEA over best baselineIDEA · Average over 5 datasetsROC-AUC: 0.8479 -> 0.8856Improved
add IDEA over best baselineIDEA · Average over 5 datasetsPrecision: 0.4050 -> 0.6022Improved
use prototypical alignment instead of direct alignment in GNN encoded spaceIDEA · Average over 5 datasetsROC-AUC: 0.7116 -> 0.8856Improved
add IDEA to ReFineReFine · Average over 5 datasetsROC-AUC: 0.6912 -> 0.7538Improved
add IDEA to ProxyExplainerProxyExplainer · Average over 5 datasetsROC-AUC: 0.8479 -> 0.8690Improved
add IDEA to V-InfoRV-InfoR · MutagenicityROC-AUC: 0.6075 -> 0.5734Hurt
add IDEA to V-InfoRV-InfoR · Average over 5 datasetsROC-AUC: 0.6683 -> 0.6696No clear effect
use Wasserstein distance instead of KL divergence in alignment (ablation)IDEA · Average over 5 datasetsROC-AUC: 3.33% average improvementImproved
add structural information disentanglement stage (ablation)IDEA · Average over 5 datasetsROC-AUC: gap of 0.0682Improved
unify representations in prototypical space instead of aligning raw representations (MSE or InfoNCE) (ablation)IDEA · Average over 5 datasetsROC-AUCImproved
use dual-stage training instead of joint optimization (ablation)IDEA · Average over 4 datasetsROC-AUC: 0.6832 -> 0.8676Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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