Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
CAD: Co-Adapting Discriminative Features for Improved Few-Shot Classification
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 |
|---|---|---|---|
| co-adapt support and query embeddings with cross-attention (ablation) | ProtoNet · miniImageNet 5-way few-shot | accuracy: 51.61±0.44 -> 59.57±0.61; 72.28±0.36 -> 84.29±0.40 | Helped |
| add query-to-support adaptation (ablation) | CAD · miniImageNet 5-way few-shot | accuracy: 56.00±0.79 -> 77.56±0.72; 85.10±0.61 -> 87.68±0.57 | Mixed |
| add support-to-query adaptation (ablation) | CAD · miniImageNet 5-way few-shot | accuracy: 76.33±0.73 -> 77.56±0.72; 80.10±0.74 -> 87.68±0.57 | Helped |
| replace cross-adaptation with self-attention within each set (ablation) | CAD · miniImageNet 5-way few-shot | accuracy: 56.01±0.86 -> 77.56±0.72; 79.94±0.72 -> 87.68±0.57 | Helped |
| use cross-relational feature adaptation | CAD · CIFAR-FS | accuracy: 79.97±0.72 vs 74.51±0.46; 94.13±0.41 vs 86.60±0.32 | Helped |
| use cross-relational feature adaptation | CAD · CUB-200-2011 | accuracy: 82.95±0.67 vs 83.55±0.19; 90.80±0.51 vs 92.92±0.10 | Mixed |
| use cross-relational feature adaptation | CAD · miniImageNet to CUB-200-2011 | accuracy: 74.10±0.75 vs 54.11±0.19; 86.37±0.60 vs 77.09±0.15 | Helped |
Closest papers by meaning
- Cross-Domain Few-Shot Classification via Learned Feature-Wise Transformation (ICLR 2020 · similarity 0.90)
- Few-Shot Classification With Feature Map Reconstruction Networks (CVPR 2021 · similarity 0.90)
- Pushing the Limit of Fine-Tuning for Few-Shot Learning: Where Feature Reusing Meets Cross-Scale Attention (AAAI 2024 · similarity 0.90)
- A Closer Look at Few-shot Classification Again (ICML 2023 · similarity 0.90)
- Few-Shot Learning via Embedding Adaptation With Set-to-Set Functions (CVPR 2020 · similarity 0.89)
- Discriminative Sample-Guided and Parameter-Efficient Feature Space Adaptation for Cross-Domain Few-Shot Learning (CVPR 2024 · 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.