Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
PARN: Position-Aware Relation Networks for Few-Shot Learning
Ziyang Wu, Yuwei Li, Lihua Guo, Kui Jia
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 |
|---|---|---|---|
| replace standard feature extractor with deformable one (DFE-4), 1-shot (ablation) | RN · Mini-ImageNet 5-way 1-shot | 5-way accuracy (%): 51.64 -> 52.07 | Improved |
| replace standard feature extractor with deformable one (DFE-4), 5-shot (ablation) | RN · Mini-ImageNet 5-way 5-shot | 5-way accuracy (%): 66.08 -> 67.53 | Improved |
| use DFE-4 instead of deeper SFE-6 with more parameters (ablation) | RN · Mini-ImageNet 5-way 1-shot | 5-way accuracy (%): 51.74 -> 52.07 | Improved |
| deepen standard extractor from 4 to 6 layers (ablation) | RN · Mini-ImageNet 5-way 1-shot | 5-way accuracy (%): 51.64 -> 51.74 | No clear effect |
| add self-correlation attention (SCA) (ablation) | RN · Mini-ImageNet 5-way 5-shot | 5-way accuracy (%): 66.00 -> 67.14 | Improved |
| add cross-correlation attention (CCA) (ablation) | RN · Mini-ImageNet 5-way 1-shot | 5-way accuracy (%): 51.29 -> 53.88 | Improved |
| add cross-correlation attention (CCA) (ablation) | RN · Mini-ImageNet 5-way 5-shot | 5-way accuracy (%): 66.00 -> 69.49 | Improved |
| add both CCA and SCA (full DCA) (ablation) | RN · Mini-ImageNet 5-way 1-shot | 5-way accuracy (%): 51.29 -> 54.36 | Improved |
| add both CCA and SCA (full DCA) (ablation) | RN · Mini-ImageNet 5-way 5-shot | 5-way accuracy (%): 66.00 -> 70.50 | Improved |
| use CCA instead of SCA (ablation) | RN with attention · Mini-ImageNet 5-way 1-shot | 5-way accuracy (%): 52.64 -> 53.88 | Improved |
| add 1x1 convolution before RN (control) (ablation) | RN · Mini-ImageNet 5-way 1-shot | 5-way accuracy (%): 51.64 -> 51.29 | No clear effect |
| add DFE and DCA together (full PARN) | RN · Mini-ImageNet 5-way 1-shot | 5-way accuracy (%): 51.64 -> 55.22 | Improved |
Closest papers by meaning
- Learning to Compare: Relation Network for Few-Shot Learning (CVPR 2018 · similarity 0.91)
- One-Shot Learning for Long-Tail Visual Relation Detection (AAAI 2020 · similarity 0.87)
- Broadcasting Convolutional Network for Visual Relational Reasoning (ECCV 2018 · similarity 0.87)
- Local Relation Networks for Image Recognition (ICCV 2019 · similarity 0.87)
- Collect and Select: Semantic Alignment Metric Learning for Few-Shot Learning (ICCV 2019 · similarity 0.87)
- Neural Snowball for Few-Shot Relation Learning (AAAI 2020 · 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.
