Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Transductive Episodic-Wise Adaptive Metric for Few-Shot Learning
Limeng Qiao, Yemin Shi, Jia Li, Yaowei Wang, Tiejun Huang, Yonghong Tian
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 TIM task-mixing augmentation to baseline (ablation) | ProtoNet baseline · miniImageNet 5-way 1-shot (ConvNet) | accuracy (%): 51.68 -> 52.97 | Improved |
| add TIM task-mixing augmentation to baseline (ablation) | ProtoNet baseline · miniImageNet 5-way 5-shot (ConvNet) | accuracy (%): 68.71 -> 70.45 | Improved |
| add episodic-wise adaptive metric (EAM) on top of TIM (ablation) | TEAM · miniImageNet 5-way 1-shot (ConvNet) | accuracy (%): 52.97 -> 55.35 | Improved |
| add EAM on top of TIM (ablation) | TEAM · CUB 5-way 1-shot (ConvNet) | accuracy (%): 70.27 -> 75.06 | Improved |
| add Bi-SIM bi-directional similarity on top of TIM+EAM (ablation) | TEAM · miniImageNet 5-way 1-shot (ConvNet) | accuracy (%): 55.35 -> 56.57 | Improved |
| add Bi-SIM on top of TIM+EAM (ablation) | TEAM · CUB 5-way 5-shot (ConvNet) | accuracy (%): 86.06 -> 86.04 | No clear effect |
| add Bi-SIM on top of TIM+EAM (ablation) | TEAM · Cifar-100 5-way 1-shot (ConvNet) | accuracy (%): 62.76 -> 64.07 | Improved |
| apply full TEAM transductive metric over the baseline | ProtoNet baseline · miniImageNet 5-way 1-shot (ConvNet) | accuracy (%): 51.68 -> 56.57 | Improved |
| apply full TEAM over the baseline | ProtoNet baseline · Cifar-100 5-way 1-shot (ConvNet) | accuracy (%): 57.83 -> 64.07 | Improved |
| apply full TEAM over the baseline | ProtoNet baseline · miniImageNet 5-way 1-shot (ResNet) | accuracy (%): 55.25 -> 60.07 | Improved |
| increase number of shots from 1 to 9 for TEAM vs baseline | TEAM · miniImageNet 5-way k-shot | accuracy (%): 4.89 -> 1.69 | Mixed |
| use TEAM-semi instead of TPN-semi in semi-supervised few-shot | TEAM-semi · miniImageNet 5-way 1-shot (semi-supervised) | accuracy (%): 52.78 -> 54.81 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.
- TADAM — TADAM: Task dependent adaptive metric for improved few-shot learning (NeurIPS 2018) · better on 0 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- TADAM: Task dependent adaptive metric for improved few-shot learning (NeurIPS 2018 · similarity 0.89)
- Finding Task-Relevant Features for Few-Shot Learning by Category Traversal (CVPR 2019 · similarity 0.89)
- Boosting Few-Shot Learning With Adaptive Margin Loss (CVPR 2020 · similarity 0.89)
- On Episodes, Prototypical Networks, and Few-Shot Learning (NeurIPS 2021 · similarity 0.89)
- Parameterless Transductive Feature Re-representation for Few-Shot Learning (ICML 2021 · similarity 0.88)
- Transductive Zero-Shot and Few-Shot CLIP (CVPR 2024 · 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.
