Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Selecting Relevant Features from a Multi-domain Representation for Few-shot Classification
Nikita Dvornik, Cordelia Schmid, Julien Mairal
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 |
|---|---|---|---|
| select features with SUR instead of concatenating all domain features (ablation) | 8 ResNet18 multi-domain set · Meta-Dataset ImageNet | accuracy: 19.5 -> 56.0 | Improved |
| select features with SUR instead of concatenating all domain features (ablation) | 8 ResNet18 multi-domain set · Meta-Dataset Aircraft | accuracy: 33.7 -> 85.7 | Improved |
| select features with SUR instead of single ImageNet extractor (ablation) | ImageNet-F · Meta-Dataset Aircraft | accuracy: 50.4 -> 85.7 | Improved |
| select features with SUR instead of single ImageNet extractor (ablation) | ImageNet-F · Meta-Dataset Fungi | accuracy: 39.1 -> 62.8 | Improved |
| select features with SUR instead of union-trained extractor (ablation) | Union-F · Meta-Dataset CIFAR 100 | accuracy: 39.5 -> 54.5 | Improved |
| select features with SUR instead of ImageNet extractor on ImageNet-like data (ablation) | ImageNet-F · Meta-Dataset CIFAR 10 | accuracy: 66.3 -> 64.6 | Hurt |
| use parametric network family (FiLM) instead of 8 separate networks | SUR · Meta-Dataset ImageNet-like vs Fungi | accuracy: ImageNet 56.3 -> 56.4; Fungi 63.1 -> 48.2 | Mixed |
| merge datasets into 3 visual domains (SUR-merge) | SUR · Meta-Dataset Aircraft | accuracy: 85.4 -> 90.1 | Improved |
| merge datasets into 3 visual domains (SUR-merge) | SUR · Meta-Dataset MNIST | accuracy: 94.3 -> 90.8 | Hurt |
| remove Birds, Textures, VGG Flower extractors (SUR 5/8) (ablation) | SUR · Meta-Dataset Traffic Sign | accuracy: 66.1 -> 63.6 | Hurt |
| remove Birds, Textures, VGG Flower extractors (SUR 5/8) (ablation) | SUR · Meta-Dataset MNIST | accuracy: 91.2 -> 92.5 | Improved |
| add SUR selection over last 6 layers | Cls ResNet12 · mini-ImageNet 5-shot | accuracy: 76.28 -> 79.25 | Improved |
Closest papers by meaning
- Cross-Domain Few-Shot Classification via Learned Feature-Wise Transformation (ICLR 2020 · similarity 0.91)
- Discriminative Sample-Guided and Parameter-Efficient Feature Space Adaptation for Cross-Domain Few-Shot Learning (CVPR 2024 · similarity 0.91)
- Few-Shot Image Classification: Just Use a Library of Pre-Trained Feature Extractors and a Simple Classifier (ICCV 2021 · similarity 0.90)
- Pushing the Limit of Fine-Tuning for Few-Shot Learning: Where Feature Reusing Meets Cross-Scale Attention (AAAI 2024 · similarity 0.89)
- Cross-domain Few-shot Classification via Invariant-content Feature Reconstruction (IJCV 2026 · similarity 0.89)
- A Closer Look at Few-shot Classification Again (ICML 2023 · 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.
