Research map / Probabilistic and scientific modeling / Variational and generative sampling
Latent Guided Sampling for Combinatorial Optimization
Sobihan Surendran, Adeline Fermanian, Sylvain Le Corff
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 multiple parallel chains over a single MCMC chain (ablation) | MCMC inference · CVRP n=100 | optimality gap: 0.701% -> 0.109% | Improved |
| add interaction between chains (ablation) | Parallel MCMC · CVRP n=100 | optimality gap: 0.109% -> -0.032% | Improved |
| add stochastic-approximation parameter updates of the decoder during inference (ablation) | Interacting MCMC · CVRP n=100 | optimality gap: -0.032% -> -0.102% | Improved |
| add the cost reweighting factor exp(-lambda C) to latent sampling (ablation) | Sampling · CVRP n=100 | optimality gap: 0.721% -> 0.109% | Improved |
| replace DE or CMA-ES inference with LGS | inference method · CVRP n=100 | optimality gap: 0.135% -> -0.102% | Improved |
| replace gradient-based Adam or SGLD inference with LGS | inference method · CVRP n=100 | optimality gap: 0.592% -> -0.102% | Improved |
| replace EAS with LGS inference | EAS · CVRP n=100 | optimality gap: 0.933% -> -0.102% | Improved |
| replace COMPASS with LGS-Net on TSP | COMPASS · TSP n=150 | optimality gap: 0.128% -> 0.081% | Improved |
| replace EAS with LGS-Net on CVRP | EAS · CVRP n=150 | optimality gap: 0.515% -> 0.343% | Improved |
| use cost-weighted loss instead of POPPY loss (ablation) | LGS-Net training · TSP/CVRP | optimality gap | Improved |
Closest papers by meaning
- Optimizing Solution-Samplers for Combinatorial Problems: The Landscape of Policy-Gradient Method (NeurIPS 2023 · similarity 0.90)
- A Diffusion Model Framework for Unsupervised Neural Combinatorial Optimization (ICML 2024 · similarity 0.88)
- Improving Generalization of Neural Combinatorial Optimization for Vehicle Routing Problems via Test-Time Projection Learning (NeurIPS 2025 · similarity 0.88)
- Preference Optimization for Combinatorial Optimization Problems (ICML 2025 · similarity 0.87)
- Fast T2T: Optimization Consistency Speeds Up Diffusion-Based Training-to-Testing Solving for Combinatorial Optimization (NeurIPS 2024 · similarity 0.86)
- Neural Solver Selection for Combinatorial Optimization (ICML 2025 · similarity 0.86)
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.
