Research map / Graph learning / GNN architectures
MAG-GNN: Reinforcement Learning Boosted Graph Neural Network
Lecheng Kong, Jiarui Feng, Hao Liu, Dacheng Tao, Yixin Chen, Muhan Zhang
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 |
|---|---|---|---|
| replace random node marking (RNM) with RL-searched node tuples (ablation) | MAG-GNN · SR25 | accuracy: 93.8 -> 100 | Improved |
| add RL search steps on one size-2 node tuple (ablation) | MAG-GNN · SR25 | accuracy | Improved |
| add RL search steps on one size-2 node tuple (ablation) | MAG-GNN · EXP and CSL | accuracy | Improved |
| pretrain on 6-CYCLES before training (ablation) | MAG-GNN · ZINC | MAE: 0.106 -> 0.096 | Improved |
| pretrain on 6-CYCLES before training (ablation) | MAG-GNN · ZINC-FULL | MAE: 0.030 -> 0.023 | Improved |
| pretrain on 6-CYCLES before training (ablation) | MAG-GNN · OGBG-MOLHIV | AUROC: 77.12 -> 78.30 | Improved |
| pretrain on 3-CYCLES before training (ablation) | MAG-GNN · OGBG-MOLHIV | AUROC: 77.12 -> 76.98 | No clear effect |
| pretrain on SR25 before training (ablation) | MAG-GNN · ZINC | MAE: 0.106 -> 0.103 | Improved |
| pretrain on SR25 before training (ablation) | MAG-GNN · ZINC-FULL | MAE: 0.030 -> 0.028 | Improved |
| use larger node tuples than size two for cycle counting | MAG-GNN vs NGNN · 6-cycles | MAE: 0.0439 -> 0.0286 | Improved |
| use MAG-GNN instead of NGNN | NGNN · QM9 U0 | MAE: 0.295 -> 0.111 | Improved |
| use MAG-GNN instead of full subgraph enumeration (I2GNN) for inference | I2GNN · ZINC-FULL | inference time (ms): 1864.1 -> 385.8 | Improved |
Later papers that compared with this method
1 later accepted paper reports MAG-GNN as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- A Flexible, Equivariant Framework for Subgraph GNNs via Graph Products and Graph Coarsening (NeurIPS 2024 · better on 2 of 3 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Efficient Subgraph GNNs by Learning Effective Selection Policies (ICLR 2024 · similarity 0.91)
- A Flexible, Equivariant Framework for Subgraph GNNs via Graph Products and Graph Coarsening (NeurIPS 2024 · similarity 0.88)
- EiG-Search: Generating Edge-Induced Subgraphs for GNN Explanation in Linear Time (ICML 2024 · similarity 0.88)
- Subgraphormer: Unifying Subgraph GNNs and Graph Transformers via Graph Products (ICML 2024 · similarity 0.88)
- How Graph Neural Networks Learn: Lessons from Training Dynamics (ICML 2024 · similarity 0.87)
- Enhancing Chess Reinforcement Learning with Graph Representation (NeurIPS 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.
