Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
URS: A Unified Neural Routing Solver for Cross-Problem Zero-Shot Generalization
Changliang Zhou, Canhong Yu, Shunyu Yao, Xi Lin, Zhenkun Wang, Yu Zhou, Qingfu Zhang
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 mixed bias module (ablation) | URS · seen variants | average gap: 6.33% -> 2.74% | Improved |
| add mixed bias module (ablation) | URS · unseen variants | average gap: 10.41% -> 7.05% | Improved |
| add WEIGHT(lambda) decoder parameter generator (ablation) | URS · seen variants | average gap: 3.84% -> 2.74% | Improved |
| add WEIGHT(lambda) decoder parameter generator (ablation) | URS · unseen variants | average gap: 8.02% -> 7.05% | Improved |
| add BIAS(lambda) attention bias generator (ablation) | URS · seen variants | average gap: 3.63% -> 2.74% | Improved |
| add BIAS(lambda) attention bias generator (ablation) | URS · unseen variants | average gap: 7.81% -> 7.05% | Improved |
| add positional identifier to UDR (ablation) | URS · unseen variants (no augmentation) | average gap: 18.12% -> 8.51% | Improved |
| add unified attribute set to UDR (ablation) | URS · seen variants (no augmentation) | average gap: 9.81% -> 4.17% | Improved |
| add unified attribute set to UDR (ablation) | URS · unseen variants (no augmentation) | average gap: 18.60% -> 8.51% | Improved |
| add node-type indicator to UDR (ablation) | URS · seen variants (no augmentation) | average gap: 5.43% -> 4.17% | Improved |
| add node-type indicator to UDR (ablation) | URS · unseen variants (no augmentation) | average gap: 18.60% -> 8.51% | Improved |
| train a single multi-task model instead of per-problem single-task models | URS · ATSP100 | optimality gap: 0.62% -> 2.26% | Hurt |
Closest papers by meaning
- SPACE: Unifying Symmetric and Asymmetric Routing Problems for Generalist Neural Solver (NeurIPS 2026 · similarity 0.91)
- MVMoE: Multi-Task Vehicle Routing Solver with Mixture-of-Experts (ICML 2024 · similarity 0.91)
- Multi-Task Vehicle Routing Solver via Mixture of Specialized Experts under State-Decomposable MDP (NeurIPS 2025 · similarity 0.90)
- CaDA: Cross-Problem Routing Solver with Constraint-Aware Dual-Attention (ICML 2025 · similarity 0.90)
- A Mixed-Curvature based Pre-training Paradigm for Multi-Task Vehicle Routing Solver (ICML 2025 · similarity 0.89)
- Memory-Enhanced Neural Solvers for Routing Problems (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.
