Research map / Architectures and efficiency / RNNs and normalization
Attention, Learn to Solve Routing Problems!
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 |
|---|---|---|---|
| Use the attention model instead of the Pointer Network | TSP20 greedy decoding · TSP20 | objective: 3.93 -> 3.85 | Helped |
| Use a greedy rollout baseline instead of an exponential baseline (ablation) | Attention Model · TSP20 | objective: 3.87 -> 3.85 | Helped |
| Use a greedy rollout baseline instead of a critic baseline (ablation) | Attention Model · TSP20 | objective: 3.87 -> 3.85 | Helped |
| Use rollout rather than exponential or critic baselines (ablation) | Pointer Network · TSP20 | objective: 3.95 -> 3.93; 3.95 -> 3.93 | Helped |
| Increase encoder layers from 0 to 1 (ablation) | Attention Model · TSP20 | objective: 4.24 -> 3.87 | Helped |
| Increase encoder layers from 1 to 2 (ablation) | Attention Model · TSP20 | objective: 3.87 -> 3.85 | Helped |
| Increase encoder layers from 2 to 3 (ablation) | Attention Model · TSP20 | objective: 3.85 -> 3.85 | No clear effect |
| Increase encoder layers to 5 or 8 (ablation) | Attention Model · TSP20 | objective: 3.85; 10.43 | Mixed |
| Sample 1280 solutions rather than use one greedy solution | Attention Model · TSP | objective: 3.85 -> 3.84; 5.80 -> 5.73; 8.12 -> 7.94 | Helped |
| Sample 1280 solutions rather than use one greedy solution | Attention Model · CVRP | objective: 6.40 -> 6.25; 10.98 -> 10.62; 16.80 -> 16.23 | Helped |
| Use AM greedy instead of RL greedy | CVRP · CVRP | objective: 6.59 -> 6.40; 11.39 -> 10.98; 17.23 -> 16.80 | Helped |
| Use AM greedy instead of RL greedy | SDVRP · SDVRP | objective: 6.51 -> 6.39; 11.32 -> 10.92; 17.12 -> 16.83 | Helped |
Closest papers by meaning
- Improving Generalization of Neural Combinatorial Optimization for Vehicle Routing Problems via Test-Time Projection Learning (NeurIPS 2025 · similarity 0.90)
- Destroy and Repair Using Hyper-Graphs for Routing (AAAI 2025 · similarity 0.89)
- CaDA: Cross-Problem Routing Solver with Constraint-Aware Dual-Attention (ICML 2025 · similarity 0.89)
- Neural Combinatorial Optimization for Robust Routing Problem with Uncertain Travel Times (NeurIPS 2024 · similarity 0.88)
- Neural Solver Selection for Combinatorial Optimization (ICML 2025 · similarity 0.88)
- GLOP: Learning Global Partition and Local Construction for Solving Large-Scale Routing Problems in Real-Time (AAAI 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.