Research map / Label-efficient and robust learning / OOD and anomaly detection
NGC: A Unified Framework for Learning With Open-World Noisy Data
Zhi-Fan Wu, Tong Wei, Jianwen Jiang, Chaojie Mao, Mingqian Tang, Yu-Feng Li
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 graph-based noise correction (ablation) | NGC · CIFAR-100 sym 80% IND noise | test accuracy (%): 55.56 -> 62.70 | Improved |
| add confidence-based selection (ablation) | NGC · CIFAR-100 sym 80% IND noise | test accuracy (%): 37.82 -> 62.70 | Improved |
| add geometry-based (graph) selection (ablation) | NGC · CIFAR-100 sym 80% IND noise | test accuracy (%): 35.16 -> 62.70 | Improved |
| add instance-level contrastive loss (ablation) | NGC · CIFAR-100 sym 80% IND noise | test accuracy (%): 51.59 -> 62.70 | Improved |
| add subgraph-level contrastive loss (ablation) | NGC · CIFAR-10 + Places-365 OOD 20k | test accuracy (%): 70.39 -> 93.67 | Improved |
| add mixup augmentation (ablation) | NGC · CIFAR-100 sym 80% IND noise | test accuracy (%): 42.59 -> 62.70 | Improved |
| add AugMix augmentation (ablation) | NGC · CIFAR-100 sym 80% IND noise | test accuracy (%): 61.75 -> 62.70 | Improved |
| add AugMix augmentation (ablation) | NGC · CIFAR-10 sym 50% noise | test accuracy (%): 94.53 -> 94.54 | No clear effect |
| add confidence-based selection (ablation) | NGC · CIFAR-10 asym 40% noise | test accuracy (%): 89.68 -> 90.55 | Improved |
| add AugMix augmentation (ablation) | NGC · WebVision-50 top-1 | test accuracy (%): 77.56 -> 79.16 | Improved |
| ensemble two networks at test time | NGC · CIFAR-10 sym 50% noise | test accuracy (%): 94.54 -> 95.09 | Improved |
| raise confidence threshold eta from 0.6 to 0.9 (ablation) | NGC · CIFAR-100 + Places-365 OOD, AUROC | AUROC (%): 72.08 -> 92.23 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DivideMix — DivideMix: Learning with Noisy Labels as Semi-supervised Learning (ICLR 2020) · better on 11 of 13 shared results
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
- Learning with Open-world Noisy Data via Class-independent Margin in Dual Representation Space (AAAI 2025 · similarity 0.88)
- ROG_PL: Robust Open-Set Graph Learning via Region-Based Prototype Learning (AAAI 2024 · similarity 0.87)
- Beyond Clean Training Data: A Versatile and Model-Agnostic Framework for Out-of-Distribution Detection with Contaminated Training Data (CVPR 2025 · similarity 0.86)
- Bridging OOD Detection and Generalization: A Graph-Theoretic View (NeurIPS 2024 · similarity 0.85)
- Empowering Graph Invariance Learning with Deep Spurious Infomax (ICML 2024 · similarity 0.85)
- Harnessing Feature Resonance under Arbitrary Target Alignment for Out-of-Distribution Node Detection (NeurIPS 2025 · similarity 0.85)
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.
