Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Task Cooperation for Semi-Supervised Few-Shot Learning
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 prediction-distribution matching between similar tasks using unlabeled pool data (ablation) | ProtoNet/TACO · MiniImageNet SS-FSL, SAC and SAI | accuracy: 42.33 -> 43.97; 44.73 -> 46.56 | Helped |
| Add prediction-distribution matching between similar tasks using unlabeled pool data | ProtoNet/TACO · MiniImageNet SS-FSL, SAC and SAI, 5-shot | accuracy | Helped |
| Use task-cooperation training with unlabeled data | ProtoNet · TieredImageNet SS-FSL, 1-shot and 5-shot | accuracy | Mixed |
| Use TACO task cooperation | ResNet-12 few-shot classifier · MiniImageNet and TieredImageNet, 1-shot and 5-shot | accuracy: 66.57 vs 63.85; 82.10 vs 79.44; 71.12 vs 69.89; 85.42 vs 84.23 | Helped |
| Use unlabeled query instances during TACO evaluation | TACO† · MiniImageNet and TieredImageNet, transductive, 1-shot and 5-shot | accuracy: 68.23 vs 67.19; 83.42 vs 80.64; 75.53 vs 73.21; 85.72 vs 84.93 | Helped |
| Increase matching weight from λ=0.1 to λ=1 (ablation) | TACO · MiniImageNet SS-FSL, SAC and SAI, 1-shot | accuracy: 43.97 -> 42.31; 46.56 -> 44.65 | Hurt |
| Set temperature to T=1 instead of T=2 (ablation) | TACO · MiniImageNet SS-FSL, SAC and SAI, 1-shot | accuracy: 43.97 -> 43.58; 46.56 -> 46.50 | No clear effect |
| Set temperature to T=4 instead of T=2 (ablation) | TACO · MiniImageNet SS-FSL, SAC and SAI, 1-shot | accuracy: 43.97 -> 43.23; 46.56 -> 46.47 | No clear effect |
| Add unlabeled-data use to Semi-ProtoNet | Semi-ProtoNet · SS-FSL, 1-shot | accuracy | No clear effect |
| Use unlabeled instances to estimate class centers | Semi-ProtoNet · SS-FSL, more than one instance per class | accuracy | Helped |
Closest papers by meaning
- Learning to Self-Train for Semi-Supervised Few-Shot Classification (NeurIPS 2019 · similarity 0.89)
- Meta-Learning for Semi-Supervised Few-Shot Classification (ICLR 2018 · similarity 0.89)
- TransMatch: A Transfer-Learning Scheme for Semi-Supervised Few-Shot Learning (CVPR 2020 · similarity 0.89)
- PLATINUM: Semi-Supervised Model Agnostic Meta-Learning using Submodular Mutual Information (ICML 2022 · similarity 0.88)
- Task Agnostic Meta-Learning for Few-Shot Learning (CVPR 2019 · similarity 0.88)
- One Meta-tuned Transformer is What You Need for Few-shot Learning (ICML 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.