Research map / Graph learning / GNN architectures
Calibrate and Boost Logical Expressiveness of GNN Over Multi-Relational and Temporal Graphs
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 graph transformation F^T (ablation) | R2-TGNN · Synthetic temporal phi2 (sum) | accuracy (%): 93.1 -> 99.0 | Helped |
| add graph transformation F^T (ablation) | R2-TGNN · Synthetic temporal phi4 (sum) | accuracy (%): 85.0 -> 98.1 | Helped |
| add graph transformation F^T (ablation) | R2-TGNN · Synthetic temporal phi3 (sum) | accuracy (%): 94.5 -> 100 | Helped |
| add graph transformation F^T (ablation) | R2-TGNN · Synthetic temporal phi1 (sum) | accuracy (%): 100 -> 100 | No clear effect |
| add global readout (R-TGNN to R2-TGNN) (ablation) | R-TGNN · Synthetic temporal phi2 (sum) | accuracy (%): 61.0 -> 93.1 | Helped |
| add graph transformation F (ablation) | R2-GNN o H · Synthetic static phi4 (sum) | accuracy (%): 81.0 -> 95.5 | Helped |
| add graph transformation F (ablation) | R2-GNN o H · Synthetic static phi2 (sum) | accuracy (%): 92.4 -> 98.8 | Helped |
| use max aggregation instead of sum (ablation) | R2-TGNN o F^T · Synthetic temporal phi4 | accuracy (%): 98.1 -> 73.4 | Hurt |
| add graph transformation F (ablation) | R2-GNN · MUTAG (sum) | accuracy (%): 85.3 -> 88.2 | Helped |
| add graph transformation F (ablation) | R2-GNN · AIFB (sum) | accuracy (%): 91.7 -> 97.2 | Helped |
| add global readout (R-GNN to R2-GNN) (ablation) | R-GNN · AIFB (sum) | accuracy (%): 91.7 -> 91.7 | No clear effect |
| add global readout (R-GNN to R2-GNN) (ablation) | R-GNN · MUTAG (sum) | accuracy (%): 76.5 -> 85.3 | Helped |
Closest papers by meaning
- The Logical Expressiveness of Graph Neural Networks (ICLR 2020 · similarity 0.94)
- Towards a Complete Logical Framework for GNN Expressiveness (ICLR 2025 · similarity 0.91)
- Recurrent Graph Neural Networks and Their Connections to Bisimulation and Logic (AAAI 2024 · similarity 0.90)
- Beyond Weisfeiler-Lehman: A Quantitative Framework for GNN Expressiveness (ICLR 2024 · similarity 0.90)
- An Empirical Study of Realized GNN Expressiveness (ICML 2024 · 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.