Research map / Graph learning / Graph self-supervised learning
Learning Transferable Graph Exploration
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 |
|---|---|---|---|
| learned GME policy instead of random exploration | GME · 6x6 maze | fraction covered: 0.329 -> 0.7160 | Helped |
| learned GME policy instead of random DFS | GME · 6x6 maze | fraction covered: 0.539 -> 0.7160 | Helped |
| use full graph (Graph) instead of current node only (Node), autoregressive, 1-layer (ablation) | GME · 6x6 maze | fraction covered: 0.6603 -> 0.7160 | Helped |
| use Graph instead of Pool (unordered set), autoregressive, 1-layer (ablation) | GME · 6x6 maze | fraction covered: 0.6671 -> 0.7160 | Helped |
| use Graph instead of Pool, non-autoregressive, 1-layer (ablation) | GME · 6x6 maze | fraction covered: 0.4151 -> 0.6634 | Helped |
| autoregressive instead of non-autoregressive history, Node policy, 2-layer (ablation) | GME · 6x6 maze | fraction covered: 0.3409 -> 0.7063 | Helped |
| autoregressive instead of non-autoregressive history, Graph policy, 2-layer (ablation) | GME · 6x6 maze | fraction covered: 0.6780 -> 0.7340 | Helped |
| 2-layer instead of 1-layer GNN node features (ablation) | GME Graph autoregressive · 6x6 maze | fraction covered: 0.7160 -> 0.7340 | Helped |
| learned GME instead of AFL with 5 distinct inputs | GME · Karel | joint coverage: 0.76 -> 0.95 | Helped |
| learned GME instead of Z3 symbolic execution, 1 input | GME · Karel | coverage: 0.837 -> 0.76 | Hurt |
| learned GME instead of random generation with 100 inputs | GME · RobustFill | joint coverage: 85% -> 93% | Helped |
| remove run-time information c_t from GnnEnc model (ablation) | GnnEnc · RobustFill | coverage | Hurt |
Closest papers by meaning
- Robustification of Online Graph Exploration Methods (AAAI 2022 · similarity 0.83)
- [Re] On Explainability of Graph Neural Networks via Subgraph Explorations (NeurIPS 2023 · similarity 0.83)
- Learning to Look Around: Intelligently Exploring Unseen Environments for Unknown Tasks (CVPR 2018 · similarity 0.82)
- GRASP: Graph Reasoning via Agentic Solving and Probing of LLMs (ICML 2026 · similarity 0.82)
- Learning with a Mole: Transferable latent spatial representations for navigation without reconstruction (ICLR 2024 · similarity 0.81)
- On Transferring Transferability: Towards a Theory for Size Generalization (NeurIPS 2025 · similarity 0.81)
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.