Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Adaptive Subspaces for Few-Shot Learning
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 class-specific subspaces instead of prototype class representations | Few-shot classification · mini-ImageNet, 5-way 1-shot | accuracy: 44.53 ± 0.76 -> 51.78 ± 0.96 | Helped |
| Add mean refinement using query-set information (ablation) | DSN · mini-ImageNet, 5-way 1-shot, ResNet-12 | accuracy: 62.64 ± 0.66 -> 64.60 ± 0.72 | Helped |
| Add mean refinement using query-set information (ablation) | DSN · tiered-ImageNet, 5-way 1-shot | accuracy: 66.22 ± 0.75 -> 67.39 ± 0.82 | Helped |
| Add mean refinement using query-set information (ablation) | DSN · CIFAR-FS, 5-way 1-shot | accuracy: 72.3 ± 0.8 -> 75.6 ± 0.9 | Helped |
| Use unlabeled samples to refine class means | Semi DSN · mini-ImageNet, semi-supervised 5-way 1-shot, without distractors | accuracy: 50.41 ± 0.31 -> 53.01 ± 0.82 | Helped |
| Add distractors to the unlabeled set (ablation) | Semi DSN · mini-ImageNet, semi-supervised 5-way 1-shot | accuracy: 53.01 ± 0.82 -> 51.01 ± 0.78 | Hurt |
| Add the discriminative term to encourage separation between class subspaces (ablation) | DSN · mini-ImageNet, 5-way 1-shot | accuracy: 50.44 ± 0.88 -> 51.78 ± 0.96 | Helped |
| Add the discriminative term to encourage separation between class subspaces (ablation) | DSN · mini-ImageNet, 5-way 5-shot | accuracy: 67.22 ± 0.69 -> 68.99 ± 0.69 | Helped |
| Vary subspace dimensionality from 2 to K − 1 (ablation) | DSN · mini-ImageNet with Conv-4 | accuracy: ± 0.5% | No clear effect |
| Use class-specific subspaces rather than prototype pooling | Few-shot classification · Open MIC, 5-way 1-shot average | accuracy: 67.85 -> 69.59 | Helped |
Closest papers by meaning
- Dynamic Few-Shot Visual Learning Without Forgetting (CVPR 2018 · similarity 0.90)
- Subspace Regularizers for Few-Shot Class Incremental Learning (ICLR 2022 · similarity 0.89)
- Meta-Learning With Differentiable Convex Optimization (CVPR 2019 · similarity 0.88)
- Discriminative Sample-Guided and Parameter-Efficient Feature Space Adaptation for Cross-Domain Few-Shot Learning (CVPR 2024 · similarity 0.87)
- Provably Improving Generalization of Few-shot models with Synthetic Data (ICML 2025 · similarity 0.86)
- Pushing the Limit of Fine-Tuning for Few-Shot Learning: Where Feature Reusing Meets Cross-Scale Attention (AAAI 2024 · similarity 0.86)
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.