Research map / Label-efficient and robust learning / Noisy labels
Noisy Correspondence Learning With Meta Similarity Correction
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 |
|---|---|---|---|
| Use MSCN rather than NCR | Cross-modal retrieval · Flickr30K 20% noise, image-to-text | R@1: 75.0 -> 77.4 | Helped |
| Use MSCN rather than NCR | Cross-modal retrieval · MS-COCO 20% noise, text-to-image | R@1: 62.5 -> 64.3 | Helped |
| Use MSCN rather than NCR | Cross-modal retrieval · Flickr30K 50% noise, image-to-text | R@1: 72.9 -> 74.4 | Helped |
| Use MSCN rather than NCR | Cross-modal retrieval · MS-COCO 50% noise, image-to-text | R@1: 74.6 -> 77.5 | Helped |
| Use MSCN rather than the strongest listed baseline, SGR-C | Cross-modal retrieval · Flickr30K 70% noise, image-to-text | R@1: 65.0 -> 69.0 | Helped |
| Use MSCN rather than the strongest listed baseline, SGR-C | Cross-modal retrieval · Flickr30K 70% noise, text-to-image | R@1: 48.1 -> 49.2 | Helped |
| Use MSCN rather than the strongest listed baseline | Cross-modal retrieval · MS-COCO 70% noise, image-to-text | R@1: 69.8 -> 74.4 | Helped |
| Use MSCN rather than NCR | Cross-modal retrieval · CC152K real noise, image-to-text | R@1: 39.5 -> 40.1 | Helped |
| Add the self-adaptive margin (ablation) | MSCN · Flickr30K 20% noise, image-to-text | R@1: 75.3 -> 77.4 | Helped |
| Add the self-adaptive margin (ablation) | MSCN · Flickr30K 20% noise, text-to-image | R@1: 58.3 -> 59.6 | Helped |
| Add data purification (ablation) | MSCN · Flickr30K 20% noise, image-to-text | R@1: 75.8 -> 77.4 | Helped |
| Add data purification (ablation) | MSCN · Flickr30K 20% noise, text-to-image | R@1: 55.8 -> 59.6 | Helped |
Closest papers by meaning
- Noisy Correspondence Rectification via Asymmetric Similarity Learning (AAAI 2025 · similarity 0.91)
- Learning to Rematch Mismatched Pairs for Robust Cross-Modal Retrieval (CVPR 2024 · similarity 0.90)
- Noisy Correspondence Learning with Self-Reinforcing Errors Mitigation (AAAI 2024 · similarity 0.90)
- Robust Noisy Correspondence Learning with Equivariant Similarity Consistency (CVPR 2024 · similarity 0.89)
- ReCon: Enhancing True Correspondence Discrimination through Relation Consistency for Robust Noisy Correspondence Learning (CVPR 2025 · similarity 0.87)
- Negative Pre-aware for Noisy Cross-Modal Matching (AAAI 2024 · 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.