Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Prototypical Networks for 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 |
|---|---|---|---|
| use prototypical classification rather than the strongest listed competing method | Omniglot few-shot classification · Omniglot | accuracy: 98.1% -> 98.8% | Helped |
| use prototypical classification rather than the strongest listed competing method | miniImageNet few-shot classification · miniImageNet 5-way 5-shot | accuracy: 60.60 ± 0.71% -> 68.20 ± 0.66% | Helped |
| use prototypical classification rather than DA-SJE | CUB-200 zero-shot classification · CUB-200 50-way 0-shot | accuracy: 50.9% -> 54.6% | Helped |
| replace cosine distance with Euclidean distance (ablation) | Prototypical Networks · miniImageNet 5-way 5-shot, 5-way training | accuracy: 51.23 ± 0.63% -> 65.77 ± 0.70% | Helped |
| increase training episode way from 5 to 20 (ablation) | Prototypical Networks · miniImageNet 5-way 5-shot | accuracy: 65.77 ± 0.70% -> 68.20 ± 0.66% | Helped |
| increase training episode way from 5 to 20 (ablation) | Prototypical Networks · miniImageNet 5-way 1-shot | accuracy: 46.14 ± 0.77% -> 48.57 ± 0.79% | Helped |
| increase training episode way from 5 to 20 (ablation) | Prototypical Networks · Omniglot 5-way 1-shot | accuracy: 97.4% -> 98.7% | Helped |
| match training shot count to test shot count (ablation) | Prototypical Networks · Omniglot 5-way 5-shot | accuracy: 99.3% -> 99.3% | No clear effect |
| fit a variance per dimension for each class (ablation) | Prototypical Networks · few-shot classification | accuracy | No clear effect |
| normalize embedded attribute prototypes to unit length (ablation) | Prototypical Networks · CUB-200 zero-shot classification | accuracy | Helped |
Closest papers by meaning
- A Closer Look at Prototype Classifier for Few-shot Image Classification (NeurIPS 2022 · similarity 0.90)
- Meta-Learning for Semi-Supervised Few-Shot Classification (ICLR 2018 · similarity 0.90)
- Revisiting Prototypical Network for Cross Domain Few-Shot Learning (CVPR 2023 · similarity 0.90)
- Simple Semantic-Aided Few-Shot Learning (CVPR 2024 · similarity 0.87)
- Provably Improving Generalization of Few-shot models with Synthetic Data (ICML 2025 · similarity 0.87)
- Scaling Few-Shot Learning for the Open World (AAAI 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.