Research map / Label-efficient and robust learning / Recognition and long-tail
Improving GANs for Long-Tailed Data through Group Spectral Regularization
Harsh Rangwani, Naman Jaswani, Tejan Karmali, Varun Jampani, R. Venkatesh Babu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 gSR | SNGAN · CIFAR-10-LT (imbalance 100) | FID: 30.62 -> 18.58 | Improved |
| add gSR | BigGAN · CIFAR-10-LT (imbalance 100) | FID: 19.55 -> 12.03 | Improved |
| add gSR | BigGAN · LSUN-LT (imbalance 1000) | FID: 45.89 -> 24.93 | Improved |
| add gSR | SNGAN · CIFAR-10-LT (imbalance 1000) | IS: 6.19 -> 5.92 | Hurt |
| add gSR at 128x128 | BigGAN · LSUN-LT 128x128 (imbalance 100) | FID: 61.63 -> 16.56 | Improved |
| add gSR | StyleGAN2-ADA · CIFAR-10-LT (imbalance 100) | FID: 71.09 -> 22.76 | Improved |
| add gSR | BigGAN · iNaturalist 2019 | FID: 6.87 -> -0.51 FID (7.42%) | Improved |
| add gSR alone without LeCam+DiffAug (ablation) | SNGAN · CIFAR-10-LT (imbalance 100) | FID: 31.73 -> 27.85 | Improved |
| add LeCam+DiffAug alone without gSR (ablation) | SNGAN · CIFAR-10-LT (imbalance 100) | FID: 31.73 -> 30.62 | No clear effect |
| add gSR on top of LeCam+DiffAug (ablation) | SNGAN · CIFAR-10-LT (imbalance 100) | FID: 30.62 -> 18.58 | Improved |
| use gSR instead of group spectral normalization (ablation) | SNGAN · CIFAR-10-LT (imbalance 100) | FID: 23.97 -> 18.58 | Improved |
| use gSR instead of gSRIP (ablation) | SNGAN · CIFAR-10-LT (imbalance 100) | FID: 23.67 -> 18.58 | Improved |
Closest papers by meaning
- Taming the Tail in Class-Conditional GANs: Knowledge Sharing via Unconditional Training at Lower Resolutions (CVPR 2024 · similarity 0.91)
- Spectral Regularization for Combating Mode Collapse in GANs (ICCV 2019 · similarity 0.88)
- Stabilizing Training of Generative Adversarial Networks through Regularization (NeurIPS 2017 · similarity 0.88)
- Long-tailed Adversarial Training with Self-Distillation (ICLR 2025 · similarity 0.88)
- Pareto GAN: Extending the Representational Power of GANs to Heavy-Tailed Distributions (ICML 2021 · similarity 0.87)
- MS$^3$D: A RG Flow-Based Regularization for GAN Training with Limited Data (ICML 2024 · similarity 0.87)
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.
