Research map / Graph learning / GNN architectures
Towards Explaining the Power of Constant-depth Graph Neural Networks for Structured Linear Programming
Qian Li, Minghui Ouyang, Tian Ding, Yuyi Wang, Qingjiang Shi, Ruoyu Sun
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Provides theoretical explanation for why constant-depth GNNs can predict optimal solutions of structured linear programs well, analyzing their expressive power on LP problems.
Problem
Empirical evidence shows shallow GNNs predict LP optimal solutions well, which is counter-intuitive because LPs are global optimization problems that would seem to need deep message passing.
Approach
Theoretical analysis showing constant-depth GNNs can represent LP solution mappings for structured LPs, explaining the counter-intuitive empirical success of shallow GNNs.
Surprising findings
- Shallow GNNs (fewer than ten layers) achieve strong performance predicting LP optimal solutions, counter-intuitive given LPs are global optimization problems
Closest papers by meaning
- On the Power of Small-size Graph Neural Networks for Linear Programming (NeurIPS 2024 · similarity 0.95)
- Learning Initial Basis Selection for Linear Programming via Duality-Inspired Tripartite Graph Representation and Comprehensive Supervision (ICML 2025 · similarity 0.90)
- Expressive Power of Graph Neural Networks for (Mixed-Integer) Quadratic Programs (ICML 2025 · similarity 0.90)
- Rethinking the Capacity of Graph Neural Networks for Branching Strategy (NeurIPS 2024 · similarity 0.89)
- Are Graph Neural Networks Optimal Approximation Algorithms? (NeurIPS 2024 · similarity 0.87)
- PDHG-Unrolled Learning-to-Optimize Method for Large-Scale Linear Programming (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.
