Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
A Closer Look at Prototype Classifier for Few-shot Image Classification
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 |
|---|---|---|---|
| Apply L2 normalization to pre-trained features before prototype classification (ablation) | Baseline-w/o-linear · miniImageNet, tieredImageNet, CIFAR-FS, FC100 (1-shot) | accuracy: 46.36±0.58 -> 59.60±1.12; 50.60 ±0.87 -> 64.24 ± 0.89; 45.91 ±0.89 -> 63.78 ±0.87; 31.60±0.61 -> 40.34±0.71 | Helped |
| Apply L2 normalization to pre-trained features before prototype classification (ablation) | Baseline-w/o-linear · miniImageNet, tieredImageNet, CIFAR-FS, FC100 (5-shot) | accuracy: 73.97±0.62 -> 77.12± 0.44; 78.10 ±0.67 -> 81.93 ±0.66; 75.81 ±0.79 -> 80.14 ±0.75; 52.50 ± 0.74 -> 56.61 ± 0.71 | Helped |
| Apply EST without L2 normalization (ablation) | Baseline-w/o-linear · miniImageNet 1-shot | accuracy: 46.36±0.58 -> 51.40± 0.83 | Helped |
| Apply EST without L2 normalization (ablation) | Baseline-w/o-linear · miniImageNet 5-shot | accuracy: 73.97±0.62 -> 73.47±0.62 | Hurt |
| Apply EST+L2-norm (ablation) | Baseline-w/o-linear · FC100 1-shot | accuracy: 31.60±0.61 -> 49.54±0.91 | Helped |
| Apply LDA+L2-norm (ablation) | Baseline-w/o-linear · FC100 5-shot | accuracy: 52.50 ± 0.74 -> 64.91 ± 0.76 | Helped |
| Use centering+L2-norm (ablation) | Baseline-w/o-linear · few-shot image classification, several 1-shot settings | accuracy | Mixed |
| Apply feature transformations that reduce norm variance and within-/between-class variance ratio (ablation) | Prototype classifier · few-shot image classification, across shot counts | accuracy | Mixed |
Closest papers by meaning
- Prototype Completion With Primitive Knowledge for Few-Shot Learning (CVPR 2021 · similarity 0.91)
- Prototypical Networks for Few-shot Learning (NeurIPS 2017 · similarity 0.90)
- Meta-Learning for Semi-Supervised Few-Shot Classification (ICLR 2018 · similarity 0.90)
- Provably Improving Generalization of Few-shot models with Synthetic Data (ICML 2025 · similarity 0.88)
- Fine-Grained Prototypes Distillation for Few-Shot Object Detection (AAAI 2024 · similarity 0.87)
- Pushing the Limit of Fine-Tuning for Few-Shot Learning: Where Feature Reusing Meets Cross-Scale Attention (AAAI 2024 · similarity 0.86)
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.