Research map / Graph learning / Knowledge and temporal graphs
Towards the Explainability of Temporal Graph Networks via Memory Backtracking and Topological Attribution
Yazheng Liu, Xi Zhang, Sihong Xie, Hui Xiong
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 |
|---|---|---|---|
| add memory backtracking tree on top of neighbor-event (topology) attribution (ablation) | MemExplainer · Temporal graph explanation, nine datasets | Fidelity_KL / Fidelity_prob | Improved |
| add topology attribution tree on top of historical-event contributions (ablation) | MemExplainer · Temporal graph explanation, nine datasets | Fidelity_KL / Fidelity_prob | Improved |
| add objective-based event selection instead of top-k by accumulated contribution (ablation) | MemExplainer · Temporal graph explanation, nine datasets | Fidelity_KL / Fidelity_prob | Improved |
| increase memory backtracking depth from 2 to 10 (ablation) | MemExplainer · tgbn-trade and tgbn-genre | Fidelity_KL / Fidelity_prob: varied depth from 2 to 10; metrics first decrease then increase | Mixed |
| increase memory backtracking depth from 2 to 10 (ablation) | MemExplainer · Enron and UCI | Fidelity_KL / Fidelity_prob: varied depth from 2 to 10 | Hurt |
| use graph attention embedding instead of graph sum, or RNN instead of GRU updater (ablation) | MemExplainer vs baseline explainers · Link prediction datasets | Fidelity_prob / Fidelity_KL | No clear effect |
| fix number of neighboring events to n=20 (ablation) | MemExplainer · Link prediction datasets | Fidelity_prob / Fidelity_KL | No clear effect |
Closest papers by meaning
- TGB 2.0: A Benchmark for Learning on Temporal Knowledge Graphs and Heterogeneous Graphs (NeurIPS 2024 · similarity 0.88)
- Future Link Prediction Without Memory or Aggregation (NeurIPS 2025 · similarity 0.88)
- CoDy: Counterfactual Explainers for Dynamic Graphs (ICML 2025 · similarity 0.88)
- Using Time-Aware Graph Neural Networks to Predict Temporal Centralities in Dynamic Graphs (NeurIPS 2024 · similarity 0.88)
- Improving Temporal Link Prediction via Temporal Walk Matrix Projection (NeurIPS 2024 · similarity 0.88)
- State Space Models on Temporal Graphs: A First-Principles Study (NeurIPS 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.
