Research map / Graph learning / Graph self-supervised learning
Transformers over Directed Acyclic 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 reachability-based attention (DAGRA) (ablation) | Transformer · ogbg-code2 | Test F1: 0.1670 -> 0.1895 | Helped |
| Add reachability-based attention (DAGRA) (ablation) | Transformer · NA | RMSE: 0.284 -> 0.253 | Helped |
| Add reachability-based attention (DAGRA) (ablation) | Transformer · Self-citation | AP: 0.573 -> 0.638 | Helped |
| Add depth positional encoding (DAGPE) (ablation) | Transformer · NA | RMSE: 0.263 -> 0.253 | Helped |
| Add depth positional encoding (DAGPE) (ablation) | Transformer · Self-citation | AP: 0.594 -> 0.638 | Helped |
| Add depth positional encoding (DAGPE) (ablation) | Transformer · ogbg-code2 | Test F1: 0.1879 -> 0.1895 | No clear effect |
| Replace DAGPE with RWPE (ablation) | DAG+Transformer · NA | RMSE: 0.253 -> 0.267 | Hurt |
| Replace DAGPE with LapPE (ablation) | DAG+Transformer · NA | RMSE: 0.253 -> 0.271 | Hurt |
| Add shortest-path distance bias (ablation) | DAG+Transformer · Self-citation | AP: 0.638 -> 0.639 | No clear effect |
| Add edge-direction bias (ablation) | DAG+Transformer · ogbg-code2 | Test F1: 0.1895 -> 0.1870 | Hurt |
| Increase receptive-field radius from k=1 to k=∞ (ablation) | DAG+TF · ogbg-code2 | Test F1 score: 0.1724 -> 0.1879 | Helped |
| Increase receptive-field radius from k=1 to k=∞ (ablation) | DAG+SAT · ogbg-code2 | Test F1 score: 0.1533 -> 0.2018 | Helped |
Closest papers by meaning
- Directed Graph Grammars for Sequence-based Learning (ICML 2025 · similarity 0.88)
- Graph Transformers Dream of Electric Flow (ICLR 2025 · similarity 0.88)
- From Theory to Practice: Rethinking Green and Martin Kernels for Unleashing Graph Transformers (ICML 2025 · similarity 0.88)
- Aligning Transformers with Weisfeiler-Leman (ICML 2024 · similarity 0.88)
- Graph Generative Pre-trained Transformer (ICML 2025 · similarity 0.88)
- Understanding Transformer Reasoning Capabilities via Graph Algorithms (NeurIPS 2024 · similarity 0.88)
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.