Research map / Graph learning / Graph algorithms and clustering
MIP-GNN: A Data-Driven Framework for Guiding Combinatorial Solvers
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 |
|---|---|---|---|
| Guide node selection using predicted variable biases | CPLEX branch-and-cut · GISP | Primal Integral: default 34.19 -> MIP-GNN 14.90 on C125.9.clq; default 129.32 -> MIP-GNN 96.31 on C250.9.clq; default 96.95 -> MIP-GNN 59 | Helped |
| Guide node selection using predicted variable biases | CPLEX branch-and-cut · FCMNF (L_n200_p0.02_c500) | Primal Integral: 137.05 -> 125.97 | Helped |
| Guide node selection using predicted variable biases | CPLEX branch-and-cut · FCMNF (L_n200_p0.02_c500) | Optimality Gap: 0.33 -> 0.35 | Hurt |
| Use bias predictions to warm-start the solver | CPLEX branch-and-cut · GISP | Best solution objective value | Mixed |
| Transfer a model trained on another GISP distribution | MIP-GNN node selection · Three GISP problem sets | Primal Integral | Mixed |
| Use variable-bias confidence to select branching variables | CPLEX branch-and-cut · GISP | Primal Integral | Mixed |
| Vary the bias-label threshold tau (ablation) | MIP-GNN EdgeConvolution (ECS) · GISP C250.9.clq | Primal Integral | Mixed |
| Replace GraphSage with GIN or EdgeConvolution, or omit error propagation (ablation) | MIP-GNN · GISP C250.9.clq | Primal Integral | No clear effect |
Closest papers by meaning
- BIPNN: Learning to Solve Binary Integer Programming via Hypergraph Neural Networks (NeurIPS 2025 · similarity 0.90)
- Expressive Power of Graph Neural Networks for (Mixed-Integer) Quadratic Programs (ICML 2025 · similarity 0.90)
- Are Graph Neural Networks Optimal Approximation Algorithms? (NeurIPS 2024 · similarity 0.89)
- COMBHelper: A Neural Approach to Reduce Search Space for Graph Combinatorial Problems (AAAI 2024 · similarity 0.89)
- Neural Solver Selection for Combinatorial Optimization (ICML 2025 · similarity 0.89)
- Rethinking the Capacity of Graph Neural Networks for Branching Strategy (NeurIPS 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.