Research map / Adversarial robustness and security / Certified and empirical robustness
Improving Adversarially Robust Few-Shot Image Classification With Generalizable Representations
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 the adversarial-aware module to adversarial training (ablation) | ResNet12 embedding model · miniImageNet 5-way 1-shot | PGD accuracy (%): 29.13 -> 30.06 | Helped |
| Add adversarial-reweighted training to adversarial training (ablation) | ResNet12 embedding model · miniImageNet 5-way 1-shot | PGD accuracy (%): 29.13 -> 30.98 | Helped |
| Add feature purification to adversarial training (ablation) | ResNet12 embedding model · miniImageNet 5-way 1-shot | PGD accuracy (%): 29.13 -> 31.24 | Helped |
| Add all three components: adversarial-aware supervision, reweighting, and feature purification (ablation) | ResNet12 embedding model · miniImageNet 5-way 1-shot | PGD accuracy (%): 29.13 -> 35.18 | Helped |
| Add adversarial-aware supervision | Ours, cross-domain miniImageNet to CIFAR-FS · CIFAR-FS 5-way 1-shot, PGD | accuracy (%): 33.05 -> 37.70 | Helped |
| Use directly trained generalizable representations instead of meta-learning baselines | Ours with ResNet12 · miniImageNet 5-way 1-shot, PGD | accuracy (%): 27.46 -> 35.18 | Helped |
| Use directly trained generalizable representations instead of meta-learning baselines | Ours with ResNet12 · CIFAR-FS 5-way 1-shot, PGD | accuracy (%): 29.55 -> 39.29 | Helped |
| Increase PGD perturbation size | Ours with ResNet12 · miniImageNet 5-way 1-shot | accuracy (%): 40.70; 35.18; 31.89; 28.05 | Hurt |
Closest papers by meaning
- Adversarially Robust Few-Shot Learning: A Meta-Learning Approach (NeurIPS 2020 · similarity 0.90)
- Adversarially Robust Few-shot Learning via Parameter Co-distillation of Similarity and Class Concept Learners (CVPR 2024 · similarity 0.89)
- Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data (ICLR 2025 · similarity 0.87)
- Few-Shot Adversarial Prompt Learning on Vision-Language Models (NeurIPS 2024 · similarity 0.86)
- Mixture of Adversarial LoRAs: Boosting Robust Generalization in Meta-Tuning (NeurIPS 2024 · similarity 0.86)
- Generate Universal Adversarial Perturbations for Few-Shot Learning (NeurIPS 2024 · similarity 0.85)
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.