Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
A Closer Look at Few-shot Classification Again
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 vanilla Finetune instead of NCC/LR/URL/TSA as adaptation algorithm | Adaptation algorithms across training methods · Meta-Dataset, 5-way 5-shot | average accuracy | Helped |
| Separate learning rates for backbone and linear head during Finetune (ablation) | Finetune adaptation · Meta-Dataset | average accuracy | Helped |
| Partial-finetune (TSA/eTT) instead of vanilla Finetune | Adaptation algorithm · Meta-Dataset, 5-way 5-shot | average accuracy | Hurt |
| Increase training classes (ImageNet) instead of samples per class | CE and PN training · ImageNet-trained, 9 Meta-Dataset datasets | test error | Helped |
| Larger training set (more classes/samples) on Omniglot and ISIC | Supervised training scale · Omniglot, ISIC | test error | Mixed |
| Increase support set size (way x shot) for Finetune relative to PN | Finetune adaptation · Meta-Dataset | relative improvement over PN | Helped |
| Use MoCo-v2 instead of CE/PN for training on mini-ImageNet-trained settings | Training algorithm (self-supervised) · Meta-Dataset | average accuracy | No clear effect |
Closest papers by meaning
- A Closer Look at Few-shot Classification (ICLR 2019 · similarity 0.92)
- Rethinking Generalization in Few-Shot Classification (NeurIPS 2022 · similarity 0.91)
- Dense Classification and Implanting for Few-Shot Learning (CVPR 2019 · similarity 0.91)
- A Theoretical Analysis of the Number of Shots in Few-Shot Learning (ICLR 2020 · similarity 0.90)
- Pushing the Limit of Fine-Tuning for Few-Shot Learning: Where Feature Reusing Meets Cross-Scale Attention (AAAI 2024 · similarity 0.89)
- Simple Semantic-Aided Few-Shot Learning (CVPR 2024 · similarity 0.89)
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.