Research map / Graph learning / GNN architectures
Graph Denoising With Framelet Regularizers
Bozitao Zhong, Ruikun Li, Xuebin Zheng, Yu Guang Wang, Junbin Gao
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 |
|---|---|---|---|
| add DoT denoising layers (ablation) | GCN · Cora (hybrid noise) | accuracy (%): 54.31 -> 64.98 | Improved |
| add DoT denoising layers (ablation) | GCN · Citeseer (hybrid noise) | accuracy (%): 34.24 -> 48.16 | Improved |
| add DoT denoising layers (ablation) | UFG-S · Wiki-CS (hybrid noise) | accuracy (%): 56.57 -> 56.71 | No clear effect |
| add DoT denoising layers (ablation) | GAT · Cora (hybrid noise) | accuracy (%): 53.25 -> 64.45 | Improved |
| add DoT denoising layers (ablation) | UFG-R · Cora (hybrid noise) | accuracy (%): 51.83 -> 63.86 | Improved |
| add edge term to node-ADMM (full DoT instead of node-only) (ablation) | GCN · Cora (hybrid noise) | accuracy (%): 63.84 -> 64.98 | Improved |
| add edge term to node-ADMM (full DoT instead of node-only) (ablation) | UFG-S · Cora (hybrid noise) | accuracy (%): 58.48 -> 61.86 | Improved |
| add node denoising to edge-only ADMM (ablation) | GCN · Cora (hybrid noise) | accuracy (%): 53.24 -> 64.98 | Improved |
| apply edge-ADMM alone versus no denoising (ablation) | GCN · Cora (hybrid noise) | accuracy (%): 54.31 -> 53.24 | Hurt |
| use DoT instead of node-TV regularizer (ablation) | GCN · Citeseer (hybrid noise) | accuracy (%): 34.71 -> 48.16 | Improved |
| use node-TV instead of DoT (ablation) | UFG-S · Wiki-CS (hybrid noise) | accuracy (%): 56.71 -> 57.31 | Improved |
Closest papers by meaning
- Is Noise Conditioning Necessary? A Unified Theory of Unconditional Graph Diffusion Models (NeurIPS 2025 · similarity 0.87)
- Joint Graph Rewiring and Feature Denoising via Spectral Resonance (ICLR 2025 · similarity 0.86)
- A General Graph Spectral Wavelet Convolution via Chebyshev Order Decomposition (ICML 2025 · similarity 0.85)
- Training Robust Graph Neural Networks by Modeling Noise Dependencies (NeurIPS 2025 · similarity 0.84)
- WaveNet: Tackling Non-stationary Graph Signals via Graph Spectral Wavelets (AAAI 2024 · similarity 0.84)
- Deep Graph Neural Networks via Posteriori-Sampling-based Node-Adaptative Residual Module (NeurIPS 2024 · similarity 0.84)
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.
