Research map / Label-efficient and robust learning / Semi-supervised learning
Semi-Supervised Few-Shot Learning via Multi-Factor Clustering
Check 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 multi-factor clustering instead of label propagation or K-means for pseudo-label acquisition (ablation) | cluster-FSL · miniImageNet and tieredImageNet, 5-way 5-shot; WRN-28-10 | accuracy: 87.99±0.37% -> 89.16±0.35%; 88.48±0.53% -> 90.18±0.43% | Helped |
| Use MFC-based support-set expansion at test time (ablation) | cluster-FSL with label propagation during fine-tuning · 5-way 1-shot; WRN-28-10 | accuracy: 78.71% -> 79.32% | Helped |
| Use MFC instead of label propagation during fine-tuning (ablation) | cluster-FSL with label propagation at test time · 5-way 1-shot; WRN-28-10 | accuracy: 78.71% -> 79.81% | Helped |
| Use MFC and label propagation jointly during fine-tuning and testing (ablation) | cluster-FSL · 5-way 1-shot; WRN-28-10 | accuracy: 78.71% -> 82.70% | Helped |
| Add RandAugment to the support set (ablation) | cluster-FSL fine-tuning · miniImageNet, 5-way 1-shot; WRN-28-10 | accuracy: 79.87±1.10% -> 80.88±1.07% | Helped |
| Add MFC-based query supervision (ablation) | cluster-FSL fine-tuning without data augmentation · miniImageNet, 5-way 1-shot; WRN-28-10 | accuracy: 79.87±1.10% -> 81.73±1.09% | Helped |
| Combine RandAugment and MFC-based query supervision (ablation) | cluster-FSL fine-tuning · miniImageNet, 5-way 1-shot; WRN-28-10 | accuracy: 79.87±1.10% -> 82.70±1.03% | Helped |
| Replace the cluster-center-only distance with reconstruction errors from a dictionary of labeled samples and centers (ablation) | MFC · miniImageNet and tieredImageNet, 5-way 5-shot; WRN-28-10 | accuracy: Label propagation: 87.99±0.37% -> MFC: 89.16±0.35%; 89.45±0.45% -> 90.18±0.43% | Helped |
| Use cluster-FSL instead of the strongest listed baseline | ResNet-12 and WRN-28-10 · miniImageNet, tieredImageNet, and CUB-200-2011, 5-way 1-shot and 5-shot | accuracy | Helped |
Closest papers by meaning
- An Embarrassingly Simple Approach to Semi-Supervised Few-Shot Learning (NeurIPS 2022 · similarity 0.89)
- Learning to Self-Train for Semi-Supervised Few-Shot Classification (NeurIPS 2019 · similarity 0.88)
- TransMatch: A Transfer-Learning Scheme for Semi-Supervised Few-Shot Learning (CVPR 2020 · similarity 0.87)
- Few-Shot Learning With Noisy Labels (CVPR 2022 · similarity 0.87)
- Multi-Label Few-Shot Image Classification via Pairwise Feature Augmentation and Flexible Prompt Learning (AAAI 2025 · similarity 0.86)
- Pushing the Limit of Fine-Tuning for Few-Shot Learning: Where Feature Reusing Meets Cross-Scale Attention (AAAI 2024 · similarity 0.84)
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.