Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
FeLMi : Few shot Learning with hard Mixup
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 |
|---|---|---|---|
| add pseudo-labeling of base examples to IER (ablation) | FeLMi · FC-100 5-way 5-shot | accuracy (%): 65.00 -> 67.92 | Helped |
| add entropy filtering of pseudo-labels (ablation) | FeLMi · FC-100 5-way 5-shot | accuracy (%): 67.92 -> 67.96 | No clear effect |
| add base-novel and novel-novel mixup (ablation) | FeLMi · FC-100 5-way 5-shot | accuracy (%): 67.96 -> 68.49 | Helped |
| add margin-based hard mixup selection (ablation) | FeLMi · FC-100 5-way 5-shot | accuracy (%): 68.49 -> 68.68 | Helped |
| add novel-novel mixup only versus no mixup (ablation) | FeLMi · FC-100 5-way 5-shot | accuracy (%): 67.92 -> 68.57 | Helped |
| add novel-novel mixup only versus no mixup (ablation) | FeLMi · CIFAR-FS 5-way 5-shot | accuracy (%): 89.37 -> 89.4 | Helped |
| add base-novel mixup with lambda_b U(0,0.2) on top of novel-novel (ablation) | FeLMi · CIFAR-FS 5-way 5-shot | accuracy (%): 89.4 -> 89.47 | Helped |
| use lambda_b B(1,1) instead of U(0,0.2) for base-novel mixup (ablation) | FeLMi · CIFAR-FS 5-way 5-shot | accuracy (%): 89.47 -> 89.26 | Hurt |
| use lambda_b B(1,1) instead of U(0,0.2) for base-novel mixup (ablation) | FeLMi · FC-100 5-way 5-shot | accuracy (%): 68.68 -> 68.49 | Hurt |
| increase number of mixup samples beyond 1000 (ablation) | FeLMi · FC-100, CIFAR-FS, miniImageNet 5-shot | accuracy | Hurt |
| change Beta alpha from 1 to 0.5 for novel-novel mixup (ablation) | FeLMi · CIFAR-FS, miniImageNet | accuracy | Mixed |
| FeLMi vs Label-Halluc | FeLMi · miniImageNet 5-way 5-shot | accuracy (%): 85.87 -> 86.08 | Helped |
Closest papers by meaning
- Pushing the Limits of Simple Pipelines for Few-Shot Learning: External Data and Fine-Tuning Make a Difference (CVPR 2022 · similarity 0.89)
- Hard-Meta-Dataset++: Towards Understanding Few-Shot Performance on Difficult Tasks (ICLR 2023 · similarity 0.88)
- Pushing the Limit of Fine-Tuning for Few-Shot Learning: Where Feature Reusing Meets Cross-Scale Attention (AAAI 2024 · similarity 0.88)
- Making Large Vision Language Models to Be Good Few-Shot Learners (AAAI 2025 · similarity 0.86)
- Discriminative Sample-Guided and Parameter-Efficient Feature Space Adaptation for Cross-Domain Few-Shot Learning (CVPR 2024 · similarity 0.86)
- Boosting Few-Shot Learning via Attentive Feature Regularization (AAAI 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.