Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Low-Shot Learning From Imaginary Data
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 |
|---|---|---|---|
| add a jointly meta-trained hallucinator to PMN (ablation) | PMN with ResNet-10 features · ImageNet low-shot, novel classes, n=1 | top-5 accuracy: 43.3 -> 45.8 | Helped |
| add a jointly meta-trained hallucinator to PMN (ablation) | PMN with ResNet-50 features · ImageNet low-shot, novel classes, n=1 | top-5 accuracy: 53.3 -> 54.7 | Helped |
| add a jointly meta-trained hallucinator to PMN (ablation) | PMN with ResNet-10 features · ImageNet low-shot, all classes, n=1 | top-5 accuracy: 55.8 -> 57.6 | Helped |
| add a jointly meta-trained hallucinator to PMN (ablation) | PMN with ResNet-50 features · ImageNet low-shot, all classes, n=1 | top-5 accuracy: 64.8 -> 65.7 | Helped |
| add a jointly meta-trained hallucinator to PN (ablation) | PN · ImageNet low-shot, novel classes | top-5 accuracy: up to 6 points | Helped |
| use one fixed noise vector rather than varying noise (ablation) | trained hallucinator with PN and PMN · ImageNet low-shot | classification accuracy | Hurt |
| replicate hallucinations during training and testing (ablation) | trained hallucinator with PN and PMN · ImageNet low-shot | classification accuracy | Hurt |
| use an untrained hallucinator instead of meta-training it (ablation) | PN and PMN · ImageNet low-shot | classification accuracy | Hurt |
| use a pretrained, fixed analogy-based hallucinator instead of meta-training it (ablation) | PN and PMN · ImageNet low-shot | classification accuracy | Hurt |
| use Gaussian feature jitter instead of the learned hallucinator (ablation) | PN and PMN · ImageNet low-shot | classification accuracy | Mixed |
| use dropout-based jitter instead of the learned hallucinator (ablation) | PN and PMN · ImageNet low-shot | classification accuracy | Mixed |
| use weighted averages of real examples instead of the learned hallucinator (ablation) | PN and PMN · ImageNet low-shot | classification accuracy | Mixed |
Closest papers by meaning
- A Dual Attention Network with Semantic Embedding for Few-Shot Learning (AAAI 2019 · similarity 0.89)
- Hallucination Improves Few-Shot Object Detection (CVPR 2021 · similarity 0.88)
- Few-Shot Learning via Saliency-Guided Hallucination of Samples (CVPR 2019 · similarity 0.88)
- Making Large Vision Language Models to Be Good Few-Shot Learners (AAAI 2025 · 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)
- One Meta-tuned Transformer is What You Need for Few-shot Learning (ICML 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.