Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Universal Representation Learning From Multiple Domains for Few-Shot Classification
Wei-Hong Li, Xialei Liu, Hakan Bilen
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 CKA+KL distillation instead of L2 feature loss (ablation) | URL · Meta-Dataset ImageNet | accuracy: 55.7 -> 58.8 | Improved |
| add CKA feature loss to KL (ablation) | URL · Meta-Dataset Textures | accuracy: 73.3 -> 77.2 | Improved |
| add CKA feature loss to KL (ablation) | URL · Meta-Dataset MNIST | accuracy: 95.3 -> 94.7 | Hurt |
| add CKA feature loss to KL (ablation) | URL · Meta-Dataset VGG Flower | accuracy: 89.6 -> 92.0 | Improved |
| use CKA instead of cosine distance (ablation) | URL · Meta-Dataset VGG Flower | accuracy: 85.0 -> 91.3 | Improved |
| add distillation from domain-specific networks versus vanilla multi-domain learning (ablation) | URL · Meta-Dataset Fungi | accuracy: 61.9 -> 68.1 | Improved |
| add distillation versus vanilla multi-domain learning (ablation) | URL · Meta-Dataset Traffic Sign | accuracy: 51.0 -> 63.3 | Improved |
| add linear feature adaptation versus nearest-centroid only (ablation) | URL · Meta-Dataset Fungi | accuracy: 66.2 -> 68.1 | Improved |
| add linear feature adaptation versus Mahalanobis distance (ablation) | URL · Meta-Dataset ImageNet | accuracy: 53.9 -> 58.8 | Improved |
| add learned adaptation versus SVM on Traffic Sign (ablation) | URL · Meta-Dataset Traffic Sign | accuracy: 65.7 -> 63.3 | Hurt |
| add feature adaptors A_theta in distillation (ablation) | URL (CKA) · Meta-Dataset Birds | accuracy: +1.7 | Improved |
| add feature adaptors A_theta in distillation (ablation) | URL (CKA) · Meta-Dataset VGG Flower | accuracy: +3.6 | Improved |
Closest papers by meaning
- A Universal Representation Transformer Layer for Few-Shot Image Classification (ICLR 2021 · similarity 0.91)
- Universal Representations: A Unified Look at Multiple Task and Domain Learning (IJCV 2023 · similarity 0.90)
- Selecting Relevant Features from a Multi-domain Representation for Few-shot Classification (ECCV 2020 · similarity 0.89)
- Unsupervised Domain Adaptation With Similarity Learning (CVPR 2018 · similarity 0.88)
- Discriminative Sample-Guided and Parameter-Efficient Feature Space Adaptation for Cross-Domain Few-Shot Learning (CVPR 2024 · similarity 0.88)
- Tackling Dimensional Collapse toward Comprehensive Universal Domain Adaptation (ICML 2025 · similarity 0.87)
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.
