Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Adversarial Feature Hallucination Networks for Few-Shot Learning
Check 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 cWGAN feature synthesis (ablation) | ResNet18 + SVM few-shot classification · Mini-ImageNet 5-way 1-shot | accuracy: 52.73 -> 57.58 | Helped |
| Add the classification regularizer (CR) to a naive generator (ablation) | Feature hallucination without cWGAN · Mini-ImageNet 5-way 1-shot | accuracy: 52.73 -> 55.65 | Helped |
| Add CR to cWGAN (ablation) | cWGAN feature hallucination · Mini-ImageNet 5-way 1-shot | accuracy: 57.58 -> 60.56 | Helped |
| Add the anti-collapse regularizer (AR) to cWGAN + CR (ablation) | AFHN · Mini-ImageNet 5-way 1-shot | accuracy: 60.56 -> 62.38 | Helped |
| Use AFHN rather than the strongest reported baseline | Few-shot classification · Mini-ImageNet 5-way 1-shot | accuracy: 61.76$\pm$0.08 -> 62.38$\pm$0.72 | Helped |
| Use AFHN rather than the strongest reported baseline | Few-shot classification · Mini-ImageNet 5-way 5-shot | accuracy: 77.59$\pm$0.12 -> 78.16$\pm$0.56 | Helped |
| Use AFHN rather than the strongest reported baseline | Few-shot classification · CUB 5-way 1-shot | accuracy: 69.33$\pm$0.22 -> 70.53$\pm$1.01 | Helped |
| Use AFHN rather than the strongest reported baseline | Few-shot classification · CUB 5-way 5-shot | accuracy: 84.10 -> 83.95$\pm$0.63 | Hurt |
| Use AFHN rather than the strongest reported baseline | Few-shot classification · CIFAR100 5-way 1-shot | accuracy: 66.70 -> 68.32$\pm$0.93 | Helped |
| Use AFHN rather than the strongest reported baseline | Few-shot classification · CIFAR100 5-way 5-shot | accuracy: 79.80 -> 81.45$\pm$0.87 | Helped |
Closest papers by meaning
- Few-Shot Learning via Saliency-Guided Hallucination of Samples (CVPR 2019 · similarity 0.88)
- Label Hallucination for Few-Shot Classification (AAAI 2022 · similarity 0.88)
- Learning To Memorize Feature Hallucination for One-Shot Image Generation (CVPR 2022 · similarity 0.87)
- Adversarially Robust Few-shot Learning via Parameter Co-distillation of Similarity and Class Concept Learners (CVPR 2024 · similarity 0.83)
- Look Closer! An Adversarial Parametric Editing Framework for Hallucination Mitigation in VLMs (AAAI 2026 · similarity 0.83)
- Beyond Token Probes: Hallucination Detection via Activation Tensors with ACT-ViT (NeurIPS 2025 · similarity 0.83)
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.