Research map / Graph learning / GNN architectures
Adaptive Diffusion in Graph Neural Networks
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 |
|---|---|---|---|
| learn diffusion time t on validation set instead of training set (ablation) | ADC heat-kernel GCN · Cora | test accuracy | Helped |
| train t per feature channel and layer instead of fixed t (no sparsification) (ablation) | GCN, ARMA, JKNet · six node classification datasets | accuracy | Helped |
| train a unique t per channel and layer rather than one t per graph or per layer (ablation) | ADC · six node classification datasets | accuracy | Helped |
| replace hand-tuned GDC with ADC | GCN, ARMA, JKNet · Cora, CiteSeer, PubMed, Coauthor CS, Amazon Computers, Amazon Photo | accuracy | Helped |
| use per-hop learned coefficients (GADC) instead of heat-kernel ADC | ADC · six node classification datasets | accuracy | No clear effect |
| increase Taylor expansion steps K (ablation) | ADC · six node classification datasets | accuracy | Helped |
| propagate before feature transformation (input channels) rather than after (ablation) | ADC + GCN · six node classification datasets | accuracy | Helped |
| vary training set size per class | ADC vs GDC vs GCN · six node classification datasets | accuracy | Helped |
Closest papers by meaning
- Learning to Approximate Adaptive Kernel Convolution on Graphs (AAAI 2024 · similarity 0.90)
- A Generalized Neural Diffusion Framework on Graphs (AAAI 2024 · similarity 0.90)
- Graph Adversarial Diffusion Convolution (ICML 2024 · similarity 0.90)
- Feature Transportation Improves Graph Neural Networks (AAAI 2024 · similarity 0.88)
- Diffusing to the Top: Boost Graph Neural Networks with Minimal Hyperparameter Tuning (ICLR 2025 · similarity 0.88)
- Generalizing CNNs to graphs with learnable neighborhood quantization (NeurIPS 2024 · similarity 0.88)
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.