Research map / Graph learning / Graph algorithms and clustering
CoCo-MILP: Inter-Variable Contrastive and Intra-Constraint Competitive MILP Solution Prediction
Tianle Pu, Jianing Li, Yingying Gao, Shixuan Liu, Zijie Geng, Haoyang Liu, Chao Chen, Changjun Fan
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 CoCo-MILP (VCL + ICC layer) to Gurobi versus plain Gurobi | Gurobi · SC, 1000s | absolute primal gap: 0.16 -> 0.01 | Improved |
| add CoCo-MILP to Gurobi | Gurobi · CA, 1000s | absolute primal gap: 295.44 -> 185.73 | Improved |
| CoCo-MILP versus ConPS | Gurobi · CA, 1000s | absolute primal gap: 208.54 -> 185.73 | Improved |
| add PS or ConPS to Gurobi on WA | Gurobi · WA, 1000s | absolute primal gap: 0.42 -> 0.42 | No clear effect |
| replace BCE with Inter-Variable Contrastive Loss (VCL) (ablation) | CoCo-MILP · SC, 1000s | objective (minimize): 125.25 -> 125.06 | Improved |
| replace BCE with VCL (ablation) | CoCo-MILP · CA, 1000s | objective (maximize): 97240.00 -> 97338.64 | Improved |
| add pairwise ranking loss to VCL (ablation) | CoCo-MILP · CA, 1000s | objective (maximize): 97272.41 -> 97338.64 | Improved |
| add multi-sample contrastive loss to VCL (ablation) | CoCo-MILP · CA, 1000s | objective (maximize): 97217.40 -> 97338.64 | Improved |
| add Intra-Constraint Competitive layer to the GNN (ablation) | CoCo-MILP · SC, 1000s | objective (minimize): 125.18 -> 125.06 | Improved |
| add ICC layer to the GNN (ablation) | CoCo-MILP · CA, 1000s | objective (maximize): 97315.96 -> 97338.64 | Improved |
| CoCo-MILP versus Gurobi on MIPLIB IIS instances | Gurobi · MIPLIB IIS subset | best objective: fast0507 174.00 -> 174.00; ramos3 233.00 -> 224.00 | Mixed |
Closest papers by meaning
- RoME: Domain-Robust Mixture-of-Experts for MILP Solution Prediction across Domains (NeurIPS 2025 · similarity 0.90)
- Apollo-MILP: An Alternating Prediction-Correction Neural Solving Framework for Mixed-Integer Linear Programming (ICLR 2025 · similarity 0.88)
- Contrastive Predict-and-Search for Mixed Integer Linear Programs (ICML 2024 · similarity 0.88)
- Towards Foundation Models for Mixed Integer Linear Programming (ICLR 2025 · similarity 0.87)
- MILP-StuDio: MILP Instance Generation via Block Structure Decomposition (NeurIPS 2024 · similarity 0.87)
- ACM-MILP: Adaptive Constraint Modification via Grouping and Selection for Hardness-Preserving MILP Instance Generation (ICML 2024 · 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.
