Research map / Graph learning / GNN architectures
Identifying and Correcting Label Noise for Robust GNNs via Influence Contradiction
Wei Ju, Wei Zhang, Siyu Yi, Zhengyang Mao, Yifan Wang, Jingyang Yuan, Zhiping Xiao, Ziyue Qiao, Ming Zhang
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 structure-level ICS (ablation) | ICGNN · Cora (uniform noise 20%) | accuracy (%): 79.4 -> 80.9 | Improved |
| add attribute-level ICS (ablation) | ICGNN · Cora (uniform noise 20%) | accuracy (%): 79.2 -> 80.9 | Improved |
| add attribute-level ICS (ablation) | ICGNN · DBLP (uniform noise 20%) | accuracy (%): 78.3 -> 80.1 | Improved |
| add noise cleaning (ablation) | ICGNN · Cora (pair noise 20%) | accuracy (%): 76.9 -> 79.4 | Improved |
| add noise cleaning (ablation) | ICGNN · DBLP (pair noise 20%) | accuracy (%): 77.1 -> 80.2 | Improved |
| add pseudo-labeling loss on unlabeled nodes (ablation) | ICGNN · Cora (pair noise 20%) | accuracy (%): 77.2 -> 79.4 | Improved |
| add pseudo-labeling loss on unlabeled nodes (ablation) | ICGNN · DBLP (uniform noise 20%) | accuracy (%): 77.5 -> 80.1 | Improved |
| use graph diffusion matrix instead of adjacency matrix (ablation) | ICGNN · DBLP (pair noise 20%) | accuracy (%): 78.7 -> 80.2 | Improved |
| use graph diffusion matrix instead of adjacency matrix (ablation) | ICGNN · Cora (uniform noise 20%) | accuracy (%): 79.6 -> 80.9 | Improved |
| use fixed alpha=0.5 instead of learnable alpha (ablation) | ICGNN · Pubmed (uniform noise 20%) | accuracy (%): 79.6 -> 80.3 | Improved |
| use fixed alpha=0.5 instead of learnable alpha (ablation) | ICGNN · DBLP (uniform noise 20%) | accuracy (%): 80.3 -> 80.1 | No clear effect |
| combine structural and attributive info (alpha=0.5) rather than either alone (ablation) | ICGNN · Coauthor CS, Amazon Photo, DBLP, Pubmed (uniform noise) | accuracy (%) | Improved |
Closest papers by meaning
- NoisyGL: A Comprehensive Benchmark for Graph Neural Networks under Label Noise (NeurIPS 2024 · similarity 0.92)
- Training Robust Graph Neural Networks by Modeling Noise Dependencies (NeurIPS 2025 · similarity 0.90)
- Mitigating Label Noise on Graphs via Topological Sample Selection (ICML 2024 · similarity 0.88)
- A Simple and Yet Fairly Effective Defense for Graph Neural Networks (AAAI 2024 · similarity 0.88)
- Multi-Label Node Classification with Label Influence Propagation (ICLR 2025 · similarity 0.88)
- Exact Certification of (Graph) Neural Networks Against Label Poisoning (ICLR 2025 · 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.
