Research map / Graph learning / Graph algorithms and clustering
Learning Combinatorial Optimization Algorithms over Graphs
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 |
|---|---|---|---|
| Use S2V-DQN instead of MVCApprox-Greedy | Minimum Vertex Cover · MemeTracker | approximation ratio: 1.2220 -> 1.0021 | Helped |
| Use S2V-DQN instead of MaxcutApprox | MAXCUT · Physics | approximation ratio: 1.2825 -> 1.0223 | Helped |
| Use S2V-DQN instead of Farthest | TSP · TSPLIB | approximation ratio: 1.0800 -> 1.0475 | Helped |
| Use S2V-DQN instead of LP | Set Covering Problem · MemeTracker SCP, 1000 test graphs | approximation ratio: 1.0009 -> 1.001 | Hurt |
| Train on 50-100-node graphs and test on 1000-1200-node graphs | S2V-DQN for MVC · BA graphs | approximation ratio: 1.0033 -> 1.0062 | Helped |
| Train on 50-100-node graphs and test on 1000-1200-node graphs | S2V-DQN for MAXCUT · BA graphs | approximation ratio: 1.0150 -> 1.0038 | Helped |
| Train on 50-100-node graphs and test on 1000-1200-node graphs | S2V-DQN for TSP · Clustered instances | approximation ratio: 1.0730 -> 1.1065 | Hurt |
| Remove the critic network and use exponential moving average performance as the actor-critic baseline (ablation) | PN-AC · The paper's optimization experiments | performance | Helped |
| Use the negative version of the designed reward for TSP with insertion (ablation) | S2V-DQN for TSP · TSP with insertion helper | performance | Helped |
Closest papers by meaning
- Exploratory Combinatorial Optimization with Reinforcement Learning (AAAI 2020 · similarity 0.92)
- Erdos Goes Neural: an Unsupervised Learning Framework for Combinatorial Optimization on Graphs (NeurIPS 2020 · similarity 0.90)
- Towards General Algorithm Discovery for Combinatorial Optimization: Learning Symbolic Branching Policy from Bipartite Graph (ICML 2024 · similarity 0.88)
- GOAL: A Generalist Combinatorial Optimization Agent Learner (ICLR 2025 · similarity 0.88)
- COMBHelper: A Neural Approach to Reduce Search Space for Graph Combinatorial Problems (AAAI 2024 · similarity 0.88)
- Preference Optimization for Combinatorial Optimization Problems (ICML 2025 · 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.