Research map / Graph learning / GNN architectures
Reconstruction for Powerful Graph Representations
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| add (n-1)-reconstruction to GIN | GIN · CSL | accuracy (%): 4.66 -> 88.66 | Helped |
| add (n-1)-reconstruction to GCN | GCN · CSL | accuracy (%): 6.66 -> 100.00 | Helped |
| add (n-1)-reconstruction to PNA | PNA · CSL | accuracy (%): 10.00 -> 100.00 | Helped |
| add reconstruction to GIN on cycle detection | GIN · 6 cycles | accuracy (%): 92.7 -> 97.35 | Helped |
| add reconstruction to GCN on 8 cycles | GCN · 8 cycles | accuracy (%): 87.14 -> 97.82 | Helped |
| add reconstruction to PNA on 8 cycles | PNA · 8 cycles | accuracy (%): 84.81 -> 99.22 | Helped |
| use small subgraph size ceil(n/2) instead of n-1 (ablation) | GIN reconstruction · CSL | accuracy (%): 88.66 -> 40.66 | Hurt |
| use small subgraph size ceil(n/2) instead of n-1 (ablation) | GCN reconstruction · 4 cycles | accuracy (%): 99.00 -> 75.19 | Hurt |
| add reconstruction to GIN | GIN · ogbg-moltox21 | ROC-AUC (%): 74.91 -> 76.84 | Helped |
| add reconstruction to GIN | GIN · ogbg-molfreesolv | RSMSE: 2.411 -> 2.117 | Helped |
| add (n-1)-reconstruction to GCN | GCN · ogbg-molfreesolv | RSMSE: 2.417 -> 2.524 | Hurt |
| add reconstruction to PNA | PNA · ogbg-molesol | RSMSE: 1.140 -> 1.723 | Hurt |
Closest papers by meaning
- Beyond Weisfeiler-Lehman: A Quantitative Framework for GNN Expressiveness (ICLR 2024 · similarity 0.92)
- A Complete Expressiveness Hierarchy for Subgraph GNNs via Subgraph Weisfeiler-Lehman Tests (ICML 2023 · similarity 0.91)
- An Empirical Study of Realized GNN Expressiveness (ICML 2024 · similarity 0.90)
- Translating Subgraphs to Nodes Makes Simple GNNs Strong and Efficient for Subgraph Representation Learning (ICML 2024 · similarity 0.90)
- Towards Bridging Generalization and Expressivity of Graph Neural Networks (ICLR 2025 · similarity 0.89)
- Bridging Theory and Practice in Link Representation with Graph Neural Networks (NeurIPS 2025 · similarity 0.89)
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.