Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Cross Attention Network for Few-shot Classification
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 a global classification head and train with joint global and nearest-neighbor classification loss (ablation) | R12-proto · miniImageNet | accuracy: 55.46 -> 61.30; 69.00 -> 76.70 | Helped |
| Add the cross-attention module (ablation) | R12-proto-ac · miniImageNet | accuracy: 61.30 -> 63.85; 76.70 -> 79.44 | Helped |
| Replace a fixed mean kernel with the adaptive meta-learner kernel (ablation) | CAN-NoML-1 · miniImageNet | accuracy: 63.55 -> 63.85; 78.88 -> 79.44 | Helped |
| Replace one shared learnable convolutional kernel with the adaptive meta-learner kernel (ablation) | CAN-NoML-2 · miniImageNet | accuracy: 63.38 -> 63.85; 79.08 -> 79.44 | Helped |
| Add transductive inference that selects confident queries to augment the support set (ablation) | CAN · miniImageNet | accuracy: 63.85 -> 67.19; 79.44 -> 80.64 | Helped |
| Add transductive inference with confident query pseudo-labels (ablation) | Matching Network · miniImageNet | accuracy: 53.52* -> 56.31; 66.20* -> 69.80 | Helped |
| Add transductive inference with confident query pseudo-labels (ablation) | Prototypical Network · miniImageNet | accuracy: 53.68* -> 55.15; 70.44* -> 71.12 | Helped |
| Add transductive inference with confident query pseudo-labels (ablation) | Relation Network · miniImageNet | accuracy: 50.65* -> 52.40; 64.18* -> 65.36 | Mixed |
Closest papers by meaning
- CAD: Co-Adapting Discriminative Features for Improved Few-Shot Classification (CVPR 2022 · similarity 0.89)
- Few-Shot Classification With Feature Map Reconstruction Networks (CVPR 2021 · similarity 0.88)
- A Closer Look at Few-shot Classification (ICLR 2019 · 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)
- A Dual Attention Network with Semantic Embedding for Few-Shot Learning (AAAI 2019 · similarity 0.88)
- Discriminative Sample-Guided and Parameter-Efficient Feature Space Adaptation for Cross-Domain Few-Shot Learning (CVPR 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.