Research map / Label-efficient and robust learning / Semi-supervised learning
LaSSL: Label-Guided Self-Training for Semi-supervised Learning
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 LaSSL instead of the strongest listed baseline | CIFAR-10, 40 labels · CIFAR-10, 40 labels | top-1 testing accuracy (%): 93.19±1.08 -> 95.07± 0.78 | Helped |
| use LaSSL instead of Dash | CIFAR-10, 250 labels · CIFAR-10, 250 labels | top-1 testing accuracy (%): 95.44±0.13 -> 95.71 ±0.46 | Helped |
| use LaSSL instead of Dash | CIFAR-100, 400 labels · CIFAR-100, 400 labels | top-1 testing accuracy (%): 55.24±0.96 -> 62.33±2.69, | Helped |
| use LaSSL instead of Dash | CIFAR-100, 2500 labels · CIFAR-100, 2500 labels | top-1 testing accuracy (%): 72.82±0.21 -> 74.67± 0.65 | Helped |
| use LaSSL instead of Dash | SVHN, 40 labels · SVHN, 40 labels | top-1 testing accuracy (%): 96.97±1.59 -> 96.91±0.52 | Hurt |
| use LaSSL instead of Dash | SVHN, 250 labels · SVHN, 250 labels | top-1 testing accuracy (%): 97.83±0.10 -> 97.85± 0.13 | No clear effect |
| use LaSSL instead of SIMPLE | Mini-ImageNet, 4000 labels · Mini-ImageNet, 4000 labels | testing accuracy (%): 49.39% -> 60.14 ± 0.26 % | Helped |
| add CACL to the vanilla method (ablation) | LaSSL on CIFAR-10 · CIFAR-10, 40 labels, 100 epochs | accuracy: 75.54 -> 85.50 | Helped |
| add BLPA to CACL (ablation) | LaSSL on CIFAR-10 · CIFAR-10, 40 labels, 100 epochs | accuracy: 85.50 -> 90.24 | Helped |
| add DA to the vanilla method (ablation) | LaSSL on CIFAR-10 · CIFAR-10, 40 labels, 100 epochs | accuracy: 75.54 -> 90.42 | Helped |
| add BLPA to CACL and DA (ablation) | LaSSL on CIFAR-10 · CIFAR-10, 40 labels, 100 epochs | accuracy: 91.11 -> 91.65 | Helped |
| include class-aware instance relations in CACL using similarity threshold 0.7 instead of 1.0 (ablation) | CACL on CIFAR-10 · CIFAR-10, 40 labels | accuracy (%): 85.17 -> 89.39 | Helped |
Closest papers by meaning
- Bi-Level Optimization for Semi-Supervised Learning with Pseudo-Labeling (AAAI 2025 · similarity 0.89)
- OwMatch: Conditional Self-Labeling with Consistency for Open-World Semi-Supervised Learning (NeurIPS 2024 · similarity 0.89)
- Keep It on a Leash: Controllable Pseudo-label Generation Towards Realistic Long-Tailed Semi-Supervised Learning (NeurIPS 2025 · similarity 0.89)
- Reduction-based Pseudo-label Generation for Instance-dependent Partial Label Learning (NeurIPS 2025 · similarity 0.88)
- Semi-supervised Multi-label Learning with Balanced Binary Angular Margin Loss (NeurIPS 2024 · similarity 0.88)
- Correlation-Induced Label Prior for Semi-Supervised Multi-Label Learning (ICML 2024 · 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.