Research map / Graph learning / Graph self-supervised learning
What Information Matters? Graph Out-of-Distribution Detection via Tri-Component Information Decomposition
Danny Wang, Ruihong Qiu, Zi Huang
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 changed | Where | Measured | Effect |
|---|---|---|---|
| replace supervised loss with information bottleneck objective (ablation) | GNNSafe · Cora | FPR95: 50.53 -> 22.20 | Improved |
| replace supervised loss with information bottleneck objective (ablation) | GNNSafe · Pubmed | FPR95: 37.71 -> 16.85 | Improved |
| replace supervised loss with IB (no energy propagation) | energy score · Cora | FPR95: 65.21 -> 46.26 | Improved |
| replace SL with IB (no energy propagation) | energy score · Twitch | AUROC: 51.24 -> 70.59 | Improved |
| add conditional independence regulariser to IB (ablation) | TIDE · Twitch | FPR95: 67.26 -> 49.91 | Improved |
| add conditional independence regulariser to IB (ablation) | TIDE · Cora | FPR95: 22.20 -> 21.56 | Improved |
| add pairwise MI minimisation (ablation) | TIDE · Twitch | FPR95: 49.91 -> 46.60 | Improved |
| add pairwise MI minimisation (ablation) | TIDE · Arxiv | FPR95: 83.86 -> 83.65 | Improved |
| add pairwise MI minimisation (ablation) | TIDE · Twitch | ID ACC: 70.33 -> 68.63 | Hurt |
| add TIDE objective to DeGEM | DeGEM · Pubmed-S | FPR95: 20.07 -> 8.44 | Improved |
| add TIDE objective to DeGEM | DeGEM · Cora-S | FPR95: 18.91 -> 15.03 | Improved |
| add TIDE objective to GOLD | GOLD · Citeseer-S | FPR95: 65.98 -> 44.97 | Improved |
Closest papers by meaning
- Structural Entropy Guided Unsupervised Graph Out-Of-Distribution Detection (AAAI 2025 · similarity 0.89)
- Redundancy-Aware Test-Time Graph Out-of-Distribution Detection (NeurIPS 2025 · similarity 0.88)
- Decoupled Graph Energy-based Model for Node Out-of-Distribution Detection on Heterophilic Graphs (ICLR 2025 · similarity 0.87)
- Graph Out-of-Distribution Detection Goes Neighborhood Shaping (ICML 2024 · similarity 0.87)
- Harnessing Feature Resonance under Arbitrary Target Alignment for Out-of-Distribution Node Detection (NeurIPS 2025 · similarity 0.87)
- Empowering Graph Invariance Learning with Deep Spurious Infomax (ICML 2024 · similarity 0.87)
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.
