Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Transductive Few-Shot Learning With Prototype-Based Label Propagation by Iterative Graph Refinement
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 |
|---|---|---|---|
| Use prototype-based label propagation with iteratively refined prototypes and graph | ResNet-12 few-shot classifier · CUB 1-shot | accuracy | Helped |
| Use prototype-based label propagation with iteratively refined prototypes and graph | ResNet-12 few-shot classifier · CIFAR-FS 1-shot | accuracy | Helped |
| Add inference data augmentation | protoLP on mini-ImageNet · mini-ImageNet 1-shot | accuracy: 70.77 ± 0.30 -> 84.35 ± 0.24 | Helped |
| Add Sinkhorn normalization with a uniform class prior (ablation) | protoLP with ResNet-12 · mini-ImageNet 1-shot | accuracy: 70.04±0.29 -> 70.77±0.30 | Helped |
| Add Sinkhorn normalization with a uniform class prior (ablation) | protoLP with WRN-28-10 · tiered-ImageNet 1-shot | accuracy: 87.91±0.25 -> 89.04±0.23 | Helped |
| Use parameterized label prediction instead of standard LP (ablation) | protoLP with ResNet-12 · mini-ImageNet 1-shot | accuracy: 69.41 -> 70.04 | Helped |
| Use parameterized label prediction instead of standard LP (ablation) | protoLP with ResNet-12 · CUB 5-shot | accuracy: 91.70 -> 91.93 | Helped |
| Use inference augmentation (ablation) | protoLP on CIFAR-FS · CIFAR-FS 1-shot | accuracy: 78.66±0.24 -> 88.22 ±0.21 | Helped |
| Use a uniform class prior in an unbalanced query setting | optimal-transport methods such as PT-MAP · mini-ImageNet unbalanced queries | accuracy: 17% | Hurt |
Closest papers by meaning
- LEARNING TO PROPAGATE LABELS: TRANSDUCTIVE PROPAGATION NETWORK FOR FEW-SHOT LEARNING (ICLR 2019 · similarity 0.90)
- Laplacian Regularized Few-Shot Learning (ICML 2020 · similarity 0.88)
- Low-Shot Learning With Large-Scale Diffusion (CVPR 2018 · similarity 0.87)
- Few-Shot Learning with Graph Neural Networks (ICLR 2018 · similarity 0.87)
- Few-Shot Learning With Noisy Labels (CVPR 2022 · similarity 0.87)
- Realistic evaluation of transductive few-shot learning (NeurIPS 2021 · 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.