Research map / Label-efficient and robust learning / Semi-supervised learning
HyperMatch: Noise-Tolerant Semi-Supervised Learning via Relaxed Contrastive Constraint
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 HyperMatch instead of FixMatch | CIFAR100 SSL · CIFAR100@400 | Top-1 accuracy: +11.86% | Helped |
| use HyperMatch instead of FixMatch | CIFAR100 SSL · CIFAR100@2500 | Top-1 accuracy: +4.74% | Helped |
| use HyperMatch instead of CCSSL | CIFAR100 SSL · CIFAR100@10000 | Top-1 accuracy: 81.09 ± 0.28 vs 80.68 ± 0.16 | Helped |
| use HyperMatch instead of CCSSL | STL10 SSL · STL10 | Top-1 accuracy: 82.98 ± 0.37 vs 80.01 ± 1.39 | Helped |
| add relaxed contrastive loss to CoMatch (ablation) | CoMatch · CIFAR100@400 | accuracy: 58.11 -> 60.85 | Helped |
| add relaxed contrastive loss to CoMatch (ablation) | CoMatch · CIFAR100@2500 | accuracy: 71.63 -> 75.2 | Helped |
| replace self-supervised contrastive loss with class-aware contrastive loss (ablation) | CIFAR100 SSL · CIFAR100@400 | accuracy: 57.73 -> 60.76 | Helped |
| replace self-supervised contrastive loss with class-aware contrastive loss (ablation) | CIFAR100 SSL · CIFAR100@2500 | accuracy: 72.85 -> 75.45 | Helped |
| add relaxed contrastive loss (ablation) | CIFAR100 SSL · CIFAR100@400 | accuracy: 57.73 -> 61.69 | Helped |
| add GMM partition to apply relaxed contrastive loss to noisy labels (ablation) | CIFAR100 SSL · CIFAR100@400 | accuracy: 61.92 -> 62.35 | Helped |
| set top-K to 2 rather than 1 (ablation) | Relaxed contrastive loss · CIFAR100@400 | accuracy: 60.51 -> 62.84 | Helped |
| set top-K to 5 rather than 2 (ablation) | Relaxed contrastive loss · CIFAR100@400 | accuracy: 62.84 -> 60.67 | Hurt |
Closest papers by meaning
- Class-Aware Contrastive Semi-Supervised Learning (CVPR 2022 · similarity 0.91)
- OwMatch: Conditional Self-Labeling with Consistency for Open-World Semi-Supervised Learning (NeurIPS 2024 · similarity 0.89)
- Continuous Contrastive Learning for Long-Tailed Semi-Supervised Recognition (NeurIPS 2024 · similarity 0.89)
- Scaling Up Semi-supervised Learning with Unconstrained Unlabelled Data (AAAI 2024 · similarity 0.89)
- Contrastive Credibility Propagation for Reliable Semi-supervised Learning (AAAI 2024 · similarity 0.88)
- RegMixMatch: Optimizing Mixup Utilization in Semi-Supervised Learning (AAAI 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.