Research map / Graph learning / GNN architectures
Learning Adaptive Neighborhoods for Graph Neural Networks
Avishkar Saha, Oscar Mendez, Chris Russell, Richard Bowden
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 |
|---|---|---|---|
| learn per-node degree distribution rather than fix global k (ablation) | DGG (GCN, Cora) · Cora | accuracy (%): 78.9 -> 83.2 (k=5 vs learned) | Improved |
| add noise to edge ranking and degree estimation (ablation) | DGG (GCN, Cora) · Cora | accuracy (%): 82.4 -> 83.2 | Improved |
| add intermediate annealed adjacency loss (ablation) | DGG-wl · Cora (GCN) | accuracy (%): 86.8 -> 87.7 | Improved |
| add PTDNet low-rank regularizer to DGG-wl (ablation) | DGG-wl · Cora (GCN) | accuracy (%): 86.8 -> 87.1 | Improved |
| combine intermediate loss with low-rank regularizer (ablation) | DGG* · Cora (GCN) | accuracy (%): 87.7 -> 87.8 | Improved |
| integrate DGG into GCN backbone | GCN · PPI | accuracy (%): 53.2 -> 77.1 (DGG-wl) | Improved |
| integrate DGG into GAT backbone on PPI | GAT · PPI | accuracy (%): 97.3 -> 97.6 | No clear effect |
| replace fixed-k kNN graph with adaptive-degree DGG | ResGCN · ModelNet40 | accuracy (%): 93.3 -> 94.4 | Improved |
| replace DGM fixed-degree graph learning with DGG on trajectory prediction | Social-STGCNN · ETH/UCY mean | FDE gain (%): 10.6% -> 23.0% | Improved |
| replace DGM with DGG on basketball trajectories | DAGNet · SportVU basketball mean | ADE gain (%): 0.3% -> 11.9% | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- NodeFormer — NodeFormer: A Scalable Graph Structure Learning Transformer for Node Classification (NeurIPS 2022) · better on 1 of 1 shared result
- SLAPS — SLAPS: Self-Supervision Improves Structure Learning for Graph Neural Networks (NeurIPS 2021) · better on 1 of 1 shared result
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
- Graph Convolutional Networks With Adaptive Neighborhood Awareness (TPAMI 2024 · similarity 0.92)
- Going Deep: Graph Convolutional Ladder-Shape Networks (AAAI 2020 · similarity 0.91)
- Deeper Insights Into Graph Convolutional Networks for Semi-Supervised Learning (AAAI 2018 · similarity 0.91)
- Generalization Guarantee of Training Graph Convolutional Networks with Graph Topology Sampling (ICML 2022 · similarity 0.91)
- L2-GCN: Layer-Wise and Learned Efficient Training of Graph Convolutional Networks (CVPR 2020 · similarity 0.91)
- Generalizing CNNs to graphs with learnable neighborhood quantization (NeurIPS 2024 · similarity 0.90)
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.
