Research map / Graph learning / GNN architectures
Boost then Convolve: Gradient Boosting Meets Graph Neural Networks
Sergei Ivanov, Liudmila Prokhorenkova
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 |
|---|---|---|---|
| Train GBDT and GNN end-to-end, fitting new trees to GNN feature-gradient updates | GAT-based BGNN · House, County, VK, Avazu | RMSE: 0.54 -> 0.5; 1.45 -> 1.26; 7.22 -> 6.95; 0.1134 -> 0.109 | Improved |
| Train GBDT and GNN end-to-end, fitting new trees to GNN feature-gradient updates | GAT-based BGNN · House_class, VK_class | accuracy: 0.625 -> 0.682; 0.577 -> 0.683 | Improved |
| Train GBDT and GNN jointly | GAT-based BGNN · SLAP, DBLP | accuracy: 0.963 -> 0.95; 0.913 -> 0.889 | Mixed |
| Train GBDT and GNN jointly | GAT-based BGNN · Wiki | RMSE: 45916 -> 49222 | Hurt |
| Train GBDT and GNN jointly | GAT-based BGNN · OGB-ArXiv | accuracy: 0.70 -> 0.67 | Hurt |
| Use end-to-end BGNN rather than pretrained-GBDT Res-GNN | GAT-based GBDT-GNN · House, County, VK, Avazu | RMSE: 0.51 -> 0.5; 1.33 -> 1.26; 7.07 -> 6.95; 0.1095 -> 0.109 | Improved |
| Use end-to-end BGNN rather than pretrained-GBDT Res-GNN | GAT-based GBDT-GNN · House_class, VK_class | accuracy: 0.625 -> 0.682; 0.603 -> 0.683 | Improved |
| Use end-to-end BGNN rather than pretrained-GBDT Res-GNN | GAT-based GBDT-GNN · SLAP, DBLP | accuracy: 0.905 -> 0.95; 0.892 -> 0.889 | Mixed |
| Use end-to-end BGNN rather than pretrained-GBDT Res-GNN | GAT-based GBDT-GNN · OGB-ArXiv | accuracy: 0.70 -> 0.67 | Hurt |
| Add GBDT preprocessing to GAT | Res-GNN · House regression | RMSE: 0.54 -> 0.51 | Improved |
| Add GBDT preprocessing to GAT | Res-GNN · Wiki regression | RMSE: 45916 -> 46747 | Hurt |
| Use BGNN instead of GAT | GAT · House regression | training time: 35 $\pm$ 2 -> 20 $\pm$ 4 | Improved |
Closest papers by meaning
- Can Classic GNNs Be Strong Baselines for Graph-level Tasks? Simple Architectures Meet Excellence (ICML 2025 · similarity 0.87)
- Diffusing to the Top: Boost Graph Neural Networks with Minimal Hyperparameter Tuning (ICLR 2025 · similarity 0.87)
- Classic GNNs are Strong Baselines: Reassessing GNNs for Node Classification (NeurIPS 2024 · similarity 0.87)
- Generalizing CNNs to graphs with learnable neighborhood quantization (NeurIPS 2024 · similarity 0.86)
- TREE-G: Decision Trees Contesting Graph Neural Networks (AAAI 2024 · similarity 0.86)
- NRGBoost: Energy-Based Generative Boosted Trees (ICLR 2025 · similarity 0.86)
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.
