Research map / Learning theory and algorithms / Combinatorial and search optimization
Learning to Cut via Hierarchical Sequence/Set Model for Efficient Mixed-Integer Programming
Jie Wang, Zhihai Wang, Xijun Li, Yufei Kuang, Zhihao Shi, Fangzhou Zhu, Mingxuan Yuan, Jia Zeng, Yongdong Zhang, Feng He Wu
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 higher-level ratio model (ablation) | HEM · CORLAT | PD integral: 1939.08 -> 1079.99 | Improved |
| add higher-level ratio model (ablation) | HEM · MIPLIB mixed neos | PD integral: 13614.29 -> 8678.76 | Improved |
| add higher-level ratio model (ablation) | HEM · Max Independent Set | Time (s): 1.88 -> 1.76 | Improved |
| add learned order (sequence model) on top of HEM-ratio-order (ablation) | HEM · CORLAT | PD integral: 1416.66 -> 1251.60 | Improved |
| add learned ratio on top of learned order (HEM versus HEM-ratio) (ablation) | HEM · CORLAT | PD integral: 1251.60 -> 1079.99 | Improved |
| add learned ratio and order to hard instances (HEM versus HEM-ratio) (ablation) | HEM · MIPLIB mixed neos | PD integral: 13659.93 -> 8678.76 | Improved |
| add learned order (HEM-ratio vs HEM-ratio-order) (ablation) | HEM · MIPLIB mixed neos | PD integral: 14026.75 -> 13659.93 | Improved |
| replace HEM with HEM++ (Set2Seq and HPPO) (ablation) | HEM · Anonymous | PD integral: 16077.15 -> 15444.70 | Improved |
| replace HEM with HEM++ (ablation) | HEM · MIPLIB mixed supportcase | PD integral: 6874.80 -> 6586.04 | Improved |
| replace HEM with HEM++ (ablation) | HEM · MIPLIB mixed neos | PD integral: 8678.76 -> 8557.46 | Improved |
| use learned cut selection (HEM) over SCIP default | SCIP 8.0.0 · Max Independent Set | Time (s): 3.88 -> 1.76 | Improved |
| select cuts with fixed-ratio NV rule instead of default | SCIP 8.0.0 · Set Covering | Time (s): 4.41 -> 9.86 | Hurt |
Closest papers by meaning
- Learning to Remove Cuts in Integer Linear Programming (ICML 2024 · similarity 0.92)
- Learning to Stop Cut Generation for Efficient Mixed-Integer Linear Programming (AAAI 2024 · similarity 0.92)
- Generalization Guarantees for Learning Score-Based Branch-and-Cut Policies in Integer Programming (NeurIPS 2025 · similarity 0.90)
- Learning Cut Generating Functions for Integer Programming (NeurIPS 2024 · similarity 0.89)
- Fast and Interpretable Mixed-Integer Linear Program Solving by Learning Model Reduction (AAAI 2025 · similarity 0.88)
- Learn2Aggregate: Supervised Generation of Chvatal-Gomory Cuts Using Graph Neural Networks (AAAI 2025 · 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.
