Research map / Label-efficient and robust learning / Recognition and long-tail
RareGAN: Generating Samples for Rare Classes
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 |
|---|---|---|---|
| Use RareGAN instead of AmpMAP | DNS amplification attack sample generation · DNS amplification attacks | fidelity: 16.60 -> 4.16 | Helped |
| Use RareGAN instead of AmpMAP | DNS amplification attack sample generation · DNS amplification attacks | diversity: 1.68% -> 98.07% | Helped |
| Use RareGAN instead of AmpMAP | DNS amplification attack sample generation · DNS amplification attacks | labeling budget: 14788089 -> 200000 | Helped |
| Use unlabeled samples for the GAN loss (ablation) | ACGAN · MNIST and CIFAR10 | fidelity-diversity tradeoff | Helped |
| Use unlabeled samples for the GAN loss (ablation) | ACGAN · DNS and packet classification | fidelity-diversity tradeoff | No clear effect |
| Add active learning (ablation) | RareGAN without active learning · DNS amplification attacks | fidelity-diversity tradeoff | Helped |
| Add weighted loss (ablation) | RareGAN without weighted loss · DNS and packet classification | fidelity-diversity tradeoff | Helped |
| Add weighted loss (ablation) | RareGAN without weighted loss · Image datasets | fidelity-diversity tradeoff | No clear effect |
| Increase rare-class loss weight from 1 to 5 (ablation) | RareGAN · DNS amplification attacks | fidelity-diversity tradeoff | Helped |
| Use 2 or 4 active-learning stages instead of the other setting (ablation) | RareGAN with weight 3.0 · DNS amplification attacks | fidelity-diversity tradeoff | No clear effect |
Closest papers by meaning
- Diverse Rare Sample Generation with Pretrained GANs (AAAI 2025 · similarity 0.90)
- Taming the Tail in Class-Conditional GANs: Knowledge Sharing via Unconditional Training at Lower Resolutions (CVPR 2024 · similarity 0.89)
- HyperGAN: A Generative Model for Diverse, Performant Neural Networks (ICML 2019 · similarity 0.89)
- Activation Maximization Generative Adversarial Nets (ICLR 2018 · similarity 0.89)
- A Unified View of cGANs with and without Classifiers (NeurIPS 2021 · similarity 0.89)
- Improved Techniques for Training GANs (NeurIPS 2016 · similarity 0.89)
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.