Research map / Continual learning / Class-incremental
Low-shot Learning via Covariance-Preserving Adversarial Augmentation Networks
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 covariance preservation to the cycle-consistent adversarial augmenter (ablation) | CP-AAN with Prototypical Networks · ImageNet LSL K=1 | top-5 accuracy (%): 42.5 -> 48.4 | Helped |
| Add covariance preservation to the cycle-consistent adversarial augmenter (ablation) | CP-AAN with Prototypical Networks · ImageNet GLSL K=1 | top-5 accuracy (%): 57.6 -> 58.5 | Helped |
| Add cycle consistency to the conditional GAN augmenter (ablation) | CP-AAN with Prototypical Networks · ImageNet LSL K=1 | top-5 accuracy (%): 38.6 -> 42.5 | Helped |
| Add cycle consistency to the conditional GAN augmenter (ablation) | CP-AAN with Prototypical Networks · ImageNet GLSL K=1 | top-5 accuracy (%): 49.4 -> 57.6 | Helped |
| Use Gaussian mixture noise instead of standard noise in cycle-GAN (ablation) | CP-AAN with Prototypical Networks · ImageNet LSL K=1 | top-5 accuracy (%): 42.5 -> 46.0 | Helped |
| Use Gaussian mixture noise instead of standard noise in cycle-GAN (ablation) | CP-AAN with Prototypical Networks · ImageNet GLSL K=1 | top-5 accuracy (%): 57.6 -> 58.0 | Helped |
| Select semantically related base classes with Neighborhood Batch Sampling rather than standard batch sampling (ablation) | CP-AAN · ImageNet low-shot recognition | accuracy | Helped |
| Use soft assignment over related base classes instead of hard top-k assignment (ablation) | CP-AAN Neighborhood Batch Sampling · ImageNet low-shot recognition | accuracy | Helped |
Closest papers by meaning
- Training Generative Adversarial Networks with Limited Data (NeurIPS 2020 · similarity 0.88)
- Feature Generating Networks for Zero-Shot Learning (CVPR 2018 · similarity 0.87)
- On Leveraging Pretrained GANs for Generation with Limited Data (ICML 2020 · similarity 0.87)
- Constructing Unrestricted Adversarial Examples with Generative Models (NeurIPS 2018 · similarity 0.87)
- Adversarially Robust Few-shot Learning via Parameter Co-distillation of Similarity and Class Concept Learners (CVPR 2024 · similarity 0.87)
- Taming the Tail in Class-Conditional GANs: Knowledge Sharing via Unconditional Training at Lower Resolutions (CVPR 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.