Research map / Label-efficient and robust learning / Recognition and long-tail
Trusted Multi-View Learning Under Noisy Supervision
Y W Zhang, Cai Xiang Xu, Han Jiang, Ziyu Guan, Wei Zhao, Xiaofei He, Murat Şensoy
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 improved training pipeline W (ablation) | TMNR · Caltech (50% noise) | accuracy (%): 72.89 -> 79.79 | Improved |
| add improved training pipeline W (ablation) | TMNR · UCI (50% noise) | accuracy (%): 88.90 -> 93.45 | Improved |
| add improved training pipeline W (ablation) | TMNR · PIE (30% noise) | accuracy (%): 73.24 -> 81.62 | Improved |
| add noisy-sample refinement R (ablation) | TMNR · UCI (50% noise) | accuracy (%): 88.90 -> 92.90 | Improved |
| add noisy-sample refinement R (ablation) | TMNR · PIE (50% noise) | accuracy (%): 59.85 -> 63.38 | Improved |
| add noisy-sample refinement R (ablation) | TMNR · Caltech (30% noise) | accuracy (%): 86.82 -> 88.62 | Improved |
| add both W and R (ablation) | TMNR · Caltech (50% noise) | accuracy (%): 72.89 -> 82.34 | Improved |
| add both W and R (ablation) | TMNR · PIE (30% noise) | accuracy (%): 73.24 -> 84.26 | Improved |
| use TMNR2 instead of TMNR on clean UCI data | TMNR · UCI (0% noise) | accuracy (%): 96.90 -> 96.20 | No clear effect |
| use TMNR2 instead of TMNR | TMNR · Caltech (50% noise) | accuracy (%): 72.89 -> 82.34 | Improved |
| use TMNR2 instead of TMNR at high noise | TMNR · UPMC Food101 (70% noise) | accuracy (%): 50.15 -> 58.98 | Improved |
| vary beta and gamma (ablation) | TMNR2 · 40% noise, all datasets | accuracy (%) | Mixed |
Closest papers by meaning
- Enhancing Testing-Time Robustness for Trusted Multi-View Classification in the Wild (CVPR 2025 · similarity 0.88)
- Navigating Conflicting Views: Harnessing Trust for Learning (ICML 2025 · similarity 0.88)
- Noisy Label Calibration for Multi-View Classification (AAAI 2025 · similarity 0.88)
- Deep Fuzzy Multi-view Learning for Reliable Classification (ICML 2025 · similarity 0.87)
- Trusted Multi-View Classification with Expert Knowledge Constraints (ICML 2025 · similarity 0.87)
- Enhancing Multi-View Classification Reliability with Adaptive Rejection (AAAI 2025 · similarity 0.86)
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.
