Research map / Graph learning / GNN architectures
Deep Ensembles for Graphs with Higher-order Dependencies
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 |
|---|---|---|---|
| Sample higher-order relatives and ensemble their predictions with independent GNNs | DGE-bag · Node classification across six datasets | micro F1: DGE-bag outperformed all other methods on five data sets and was second to DGE-pool* on Air | Helped |
| Sample higher-order relatives and ensemble their predictions with independent GNNs | DGE-bag · Link prediction across five datasets | AUPRC | Mixed |
| Use separate parameters for each base GNN instead of sharing them (ablation) | DGE-bag · Node classification across datasets | micro F1 | Helped |
| Use separate parameters for each base GNN instead of sharing them (ablation) | DGE-bag · Link prediction across datasets | AUPRC | Helped |
| Use relative sampling and pooling rather than training the ensemble on the first-order graph alone (ablation) | DGE-bag · Node classification across four reported datasets | micro F1 | Helped |
| Concatenate relative representations instead of averaging their class probabilities (ablation) | DGE · Node classification across datasets | micro F1 | Hurt |
| Replace independent base learners with shared parameters (ablation) | DGE · Node classification across datasets | micro F1 | Hurt |
| Use shared-parameter pooling (DGE-pool*) (ablation) | DGE · Air node classification | micro F1: 0.856 -> 0.865 | Helped |
| Use a single shared-parameter model that resamples relatives per batch (DGE-batch*) (ablation) | DGE · Node classification across datasets | micro F1 | Hurt |
| Pool sampled-relative representations using attention (ablation) | DGE · Node classification | micro F1 | Hurt |
| Increase model depth for DGE-bag (ablation) | DGE-bag · Node classification across datasets | test loss | Mixed |
Closest papers by meaning
- Fast and Deep Graph Neural Networks (AAAI 2020 · similarity 0.89)
- Generalizing CNNs to graphs with learnable neighborhood quantization (NeurIPS 2024 · similarity 0.88)
- A Manifold Perspective on the Statistical Generalization of Graph Neural Networks (ICML 2025 · similarity 0.88)
- Mixture of Weak and Strong Experts on Graphs (ICLR 2024 · similarity 0.88)
- Sketch-Augmented Features Improve Learning Long-Range Dependencies in Graph Neural Networks (NeurIPS 2025 · similarity 0.88)
- Deep Graph Neural Networks via Posteriori-Sampling-based Node-Adaptative Residual Module (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.