Research map / Label-efficient and robust learning / Semi-supervised learning
Towards Realistic Semi-Supervised Learning
Mamshad Nayeem Rizve, Navid Kardan, Mubarak Shah
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 mixed soft/hard pseudo-labeling (ablation) | TRSSL · CIFAR-10 | Novel accuracy: 86.6 -> 89.6 | Improved |
| add mixed soft/hard pseudo-labeling (ablation) | TRSSL · CIFAR-100 | Novel accuracy: 46.6 -> 52.1 | Improved |
| add mixed soft/hard pseudo-labeling (ablation) | TRSSL · Tiny ImageNet | Novel accuracy: 16.0 -> 20.5 | Improved |
| add uncertainty-guided temperature scaling (ablation) | TRSSL · CIFAR-100 | All accuracy: 57.5 -> 60.3 | Improved |
| add uncertainty-guided temperature scaling (ablation) | TRSSL · CIFAR-10 | All accuracy: 91.9 -> 92.2 | Improved |
| add uncertainty-guided temperature scaling (ablation) | TRSSL · CIFAR-100 | Novel accuracy: 48.1 -> 52.1 | Improved |
| add both temperature scaling and mixed pseudo-labeling (ablation) | TRSSL · CIFAR-100 | All accuracy: 57.9 -> 60.3 | Improved |
| use oracle number of novel classes instead of estimated (ablation) | TRSSL · CIFAR-100 | All accuracy: 54.8 -> 60.3 | Improved |
| use imbalanced class distribution prior instead of balanced (ablation) | TRSSL · CIFAR-100 (IF=10) | All accuracy: 38.9 -> 41.0 | Improved |
| use imbalanced class distribution prior instead of balanced (ablation) | TRSSL · CIFAR-100 (IF=20) | All accuracy: 33.8 -> 36.9 | Improved |
| estimate class prior iteratively instead of balanced prior (ablation) | TRSSL · CIFAR-100 (IF=20) | All accuracy: 33.8 -> 35.3 | Improved |
| add sample reassignment to novel class number estimation (ablation) | k-means on SimCLR features · Tiny ImageNet | Estimated number of classes: 132 -> 192 | Improved |
Closest papers by meaning
- Open-World Semi-Supervised Learning (ICLR 2022 · similarity 0.92)
- Realistic Evaluation of Deep Semi-Supervised Learning Algorithms (NeurIPS 2018 · similarity 0.91)
- OpenLDN: Learning to Discover Novel Classes for Open-World Semi-Supervised Learning (ECCV 2022 · similarity 0.91)
- Class-Aware Contrastive Semi-Supervised Learning (CVPR 2022 · similarity 0.91)
- Revisiting Semi-Supervised Learning in the Era of Foundation Models (NeurIPS 2025 · similarity 0.90)
- OwMatch: Conditional Self-Labeling with Consistency for Open-World Semi-Supervised Learning (NeurIPS 2024 · similarity 0.89)
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.
