Research map / Label-efficient and robust learning / Noisy labels
Twin Contrastive Learning With Noisy Labels
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 |
|---|---|---|---|
| remove entropy regularization loss L_reg (ablation) | TCL · CIFAR-10 Sym. 90% | accuracy: 89.4 -> 34.5 without L_reg | Helped |
| remove mixup alignment loss L_align (ablation) | TCL · CIFAR-10 Sym. 90% | accuracy: 89.4 -> 84.6 without L_align | Helped |
| add noisy label detection and bootstrap cross-supervision (vs. baseline using noisy labels directly) (ablation) | TCL · CIFAR-10 Sym. 90% | accuracy: 20.6 -> 89.4 | Helped |
| use OOD-based label noise detection instead of kNN or cross-entropy detection (ablation) | label noise detection · CIFAR-10 Sym. 90% | accuracy: OOD 82.1 vs kNN 79.7 vs CE loss 75.9 | Helped |
| bootstrap cross-supervision instead of temporal ensembling (ablation) | target estimation · CIFAR-10 Sym. 90% | accuracy: 89.4 vs 72.7 (Ensem.) | Helped |
| implement TCL on MoCo framework instead of SimCLR-style contrastive framework (ablation) | TCL · CIFAR-10 Sym. 90% | accuracy: 89.4 -> 90.7 | Helped |
| TCL outperforms prior SOTA at extreme noise | TCL · CIFAR-10 Sym. 90% | accuracy: 89.4 vs 81.9 (Sel-CL+) | Helped |
| competitive or slightly lower at low noise, larger gains at extreme noise | TCL · CIFAR-10 Sym. 20% | accuracy: 95.0 vs RRL 95.8 and Sel-CL+ 95.5 | No clear effect |
| fine-tune on detected clean samples with mixup for 70 epochs (TCL+) | TCL · CIFAR-10 Sym. 50% / low noise | accuracy: 93.9 -> 94.5 (50% sym) | Helped |
Closest papers by meaning
- Adaptive Integration of Partial Label Learning and Negative Learning for Enhanced Noisy Label Learning (AAAI 2024 · similarity 0.87)
- Continuous Contrastive Learning for Long-Tailed Semi-Supervised Recognition (NeurIPS 2024 · similarity 0.86)
- Unlocking the Power of Open Set: A New Perspective for Open-Set Noisy Label Learning (AAAI 2024 · similarity 0.86)
- Learning with Open-world Noisy Data via Class-independent Margin in Dual Representation Space (AAAI 2025 · similarity 0.86)
- Label-Noise Robust Diffusion Models (ICLR 2024 · similarity 0.85)
- Learning the Latent Causal Structure for Modeling Label Noise (NeurIPS 2024 · 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.