Research map / Graph learning / Knowledge and temporal graphs
Recurrent Temporal Revision Graph Networks
YIZHOU CHEN, Anxiang Zeng, Qingtao Yu, Kerui Zhang, Cao Yuanpeng, Kangle Wu, Guangda Huzhang, Han Yu, Zhiming Zhou
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 standard temporal neighbor aggregation with recurrent temporal revision | RTRGN · MovieLens, Wikipedia, UCI-FSN, and Ecommerce (2-layer link prediction) | AP (%): 94.90\pm0.4 vs 85.33\pm0.5; 98.79\pm0.2 vs 98.38\pm0.2; 98.04\pm0.1 vs 93.48\pm0.4; 95.28\pm0.1 vs 85.59\pm0.2 | Improved |
| replace standard temporal neighbor aggregation with recurrent temporal revision | RTRGN · MovieLens, Wikipedia, UCI-FSN, and Ecommerce (2-layer link prediction) | AP (%): 93.37\pm0.3 vs 83.43\pm0.5; 98.28\pm0.2 vs 97.85\pm0.2; 95.62\pm0.1 vs 83.58\pm0.3; 92.87\pm0.1 vs 84.31\pm0.2 | Improved |
| add heterogeneous revision (HR) (ablation) | RTRGN · Oscillating CSL | Accuracy: 16.67 -> 100.0 | Improved |
| add heterogeneous revision (HR) (ablation) | RTRGN · Cuneiform Writing | Accuracy: 96.43 -> 100.0 | Improved |
| remove the retained hidden state h_u^k(t^-) (ablation) | RTRGN · MovieLens, Wikipedia, and UCI | AUC | Hurt |
| remove neighbor revision r_u^k(t), leaving a pure RNN (ablation) | RTRGN · MovieLens, Wikipedia, and UCI | AUC | Hurt |
| remove neighbor state h_v^{k-1}(t) from temporal revision (ablation) | RTRGN · MovieLens, Wikipedia, and UCI | AUC | No clear effect |
| remove neighbor state change Δh_v^{k-1}(t) from temporal revision (ablation) | RTRGN · MovieLens, Wikipedia, and UCI | AUC | Hurt |
| remove lower-level neighbor revisions r_v^{k-1}(t) from temporal revision (ablation) | RTRGN · MovieLens, Wikipedia, and UCI | AUC | Hurt |
| replace GRU for FUNC_4 with a 2-layer MLP (ablation) | RTRGN · 7 of 8 tested link prediction benchmarks | AP | Hurt |
| replace the GRU state transition with weighted averaging (ablation) | RTRGN · link prediction benchmarks | performance | Hurt |
| use an RNN cell instead of a GRU for the state transition (ablation) | RTRGN · link prediction benchmarks | performance | No clear effect |
Closest papers by meaning
- Using Time-Aware Graph Neural Networks to Predict Temporal Centralities in Dynamic Graphs (NeurIPS 2024 · similarity 0.87)
- Simple and Efficient Heterogeneous Temporal Graph Neural Network (NeurIPS 2025 · similarity 0.85)
- Improving Temporal Link Prediction via Temporal Walk Matrix Projection (NeurIPS 2024 · similarity 0.85)
- TGB 2.0: A Benchmark for Learning on Temporal Knowledge Graphs and Heterogeneous Graphs (NeurIPS 2024 · similarity 0.85)
- Temporal-Aware Evaluation and Learning for Temporal Graph Neural Networks (AAAI 2025 · similarity 0.85)
- Locality-Aware Graph Rewiring in GNNs (ICLR 2024 · similarity 0.85)
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.
