Research map / Reinforcement learning / Visual and representation RL
Reinforcement Learning for Solving the Vehicle Routing Problem
MohammadReza Nazari, Afshin Oroojlooy, Lawrence Snyder, Martin Takac
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 |
|---|---|---|---|
| Use beam search instead of greedy decoding (ablation) | RL · VRP10, Cap20 | mean tour length: 4.84 -> 4.68 | Improved |
| Use beam search instead of greedy decoding (ablation) | RL · VRP20, Cap30 | mean tour length: 6.59 -> 6.40 | Improved |
| Use beam search instead of greedy decoding (ablation) | RL · VRP50, Cap40 | mean tour length: 11.39 -> 11.15 | Improved |
| Use beam search instead of greedy decoding (ablation) | RL · VRP100, Cap50 | mean tour length: 17.23 -> 16.96 | Improved |
| Allow split deliveries by relaxing the demand-versus-load mask (ablation) | RL-BS(10) · VRP10, Cap20 | mean tour length: 4.68 -> 4.65 | Improved |
| Allow split deliveries by relaxing the demand-versus-load mask (ablation) | RL-BS(10) · VRP20, Cap30 | mean tour length: 6.40 -> 6.34 | Improved |
| Allow split deliveries by relaxing the demand-versus-load mask (ablation) | RL-BS(10) · VRP50, Cap40 | mean tour length: 11.15 -> 11.08 | Improved |
| Allow split deliveries by relaxing the demand-versus-load mask (ablation) | RL-BS(10) · VRP100, Cap50 | mean tour length: 16.96 -> 16.86 | Improved |
| Remove the embedding layer (ablation) | RL · VRP | solution quality | Hurt |
| Add dynamic elements as inputs to the decoder (ablation) | RL · VRP | solution quality | No clear effect |
| Add an entropy regularizer (ablation) | RL training · VRP | solution quality | No clear effect |
| Remove the RNN encoder (ablation) | Pointer-RL · TSP20 | average tour length: 3.96 -> 3.97 | No clear effect |
Closest papers by meaning
- DPN: Decoupling Partition and Navigation for Neural Solvers of Min-max Vehicle Routing Problems (ICML 2024 · similarity 0.89)
- Memory-Enhanced Neural Solvers for Routing Problems (NeurIPS 2025 · similarity 0.89)
- Preference Optimization for Combinatorial Optimization Problems (ICML 2025 · similarity 0.88)
- Multi-Task Vehicle Routing Solver via Mixture of Specialized Experts under State-Decomposable MDP (NeurIPS 2025 · similarity 0.88)
- MVMoE: Multi-Task Vehicle Routing Solver with Mixture-of-Experts (ICML 2024 · similarity 0.87)
- MTL-KD: Multi-Task Learning Via Knowledge Distillation for Generalizable Neural Vehicle Routing Solver (NeurIPS 2025 · 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.
