Research map / Label-efficient and robust learning / Self-supervised and contrastive
Boosting Few-Shot Visual Learning With Self-Supervision
Spyros Gidaris, Andrei Bursuc, Nikos Komodakis, Patrick Perez, Matthieu Cord
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 rotation prediction self-supervision (ablation) | Cosine Classifier (WRN-28-10) · MiniImageNet 5-way 1-shot | accuracy (%): 61.09 -> 62.93 | Improved |
| add rotation prediction self-supervision (ablation) | Prototypical Networks (WRN-28-10) · MiniImageNet 5-way 5-shot | accuracy (%): 68.72 -> 72.13 | Improved |
| add rotation prediction self-supervision (ablation) | Cosine Classifier (WRN-28-10) · CIFAR-FS 5-way 1-shot | accuracy (%): 71.83 -> 73.62 | Improved |
| add rotation prediction self-supervision (ablation) | Prototypical Networks (WRN-28-10) · CIFAR-FS 5-way 5-shot | accuracy (%): 81.79 -> 81.25 | Hurt |
| add rotation prediction self-supervision (ablation) | Cosine Classifier (WRN-28-10) · tiered-MiniImageNet 5-way 1-shot | accuracy (%): 70.04 -> 70.53 | Improved |
| add relative patch location self-supervision (ablation) | Cosine Classifier (WRN-28-10) · MiniImageNet 5-way 1-shot | accuracy (%): 58.43 -> 60.71 | Improved |
| add rotation augmentation to CC baseline (ablation) | Cosine Classifier (Conv-4-64) · MiniImageNet 5-way 5-shot | accuracy (%): 71.13 -> 70.89 | Hurt |
| add rotation augmentation to CC baseline (ablation) | Cosine Classifier (WRN-28-10) · MiniImageNet 5-way 1-shot | accuracy (%): 58.59 -> 61.09 | Improved |
| add auxiliary patch classification loss to CC baseline (ablation) | Cosine Classifier (WRN-28-10) · MiniImageNet 5-way 1-shot | accuracy (%): 58.59 -> 58.43 | Hurt |
| add unlabeled MiniImageNet images with self-supervision (5% labels) (ablation) | CC+rot (WRN-28-10) · MiniImageNet 5% labels 5-way 1-shot | accuracy (%): 43.60 -> 47.25 | Improved |
| add unlabeled tiered-MiniImageNet images (5% labels) (ablation) | CC+rot (WRN-28-10) · MiniImageNet 5% labels 5-way 1-shot | accuracy (%): 43.60 -> 46.95 | Improved |
| add unlabeled tiered-MiniImageNet images (full labels) | CC+rot (WRN-28-10) · MiniImageNet 5-way 1-shot | accuracy (%): 62.93 -> 63.77 | Improved |
Closest papers by meaning
- Self-Supervision Can Be a Good Few-Shot Learner (ECCV 2022 · similarity 0.93)
- LibFewShot: A Comprehensive Library for Few-Shot Learning (TPAMI 2023 · similarity 0.91)
- Pushing the Limits of Simple Pipelines for Few-Shot Learning: External Data and Fine-Tuning Make a Difference (CVPR 2022 · similarity 0.91)
- Rethinking Few-shot Image Classification: A Good Embedding is All You Need? (ECCV 2020 · similarity 0.90)
- Self-training For Few-shot Transfer Across Extreme Task Differences (ICLR 2021 · similarity 0.90)
- Learning to Self-Train for Semi-Supervised Few-Shot Classification (NeurIPS 2019 · similarity 0.90)
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.
