Research map / Image generation / GANs
KD-DLGAN: Data Limited Image Generation via Knowledge Distillation
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 KD-DLGAN to DA | BigGAN · CIFAR-10, 10% data | FID: 23.34 \textpm \ 0.09 -> 14.20 \textpm \ 0.06 | Helped |
| Add KD-DLGAN to DA | BigGAN · CIFAR-100, 10% data | FID: 35.39 \textpm \ 0.08 -> 18.03 \textpm \ 0.11 | Helped |
| Add KD-DLGAN to DA | BigGAN · ImageNet, 10% training data | FID: 32.82 \textpm \ 0.18 -> 19.99 \textpm \ 0.11 | Helped |
| Add KD-DLGAN to DA | StyleGAN-v2 · 100-shot Obama | FID: 46.87 -> 31.54 \textpm \ 0.27 | Helped |
| Add AGKD (ablation) | DA with BigGAN · CIFAR-10, 10% data | FID: 23.34 -> 15.85 \textpm \ 0.06 | Helped |
| Add CGKD (ablation) | DA with BigGAN · CIFAR-10, 10% data | FID: 23.34 -> 18.66 \textpm \ 0.09 | Helped |
| Combine AGKD and CGKD rather than use either alone (ablation) | DA with BigGAN · CIFAR-10, 10% data | FID: 15.85 \textpm \ 0.06 and 18.66 \textpm \ 0.09 -> 14.20 \textpm \ 0.06 | Helped |
| Add AGKD (ablation) | DA with StyleGAN-v2 · 100-shot Obama | FID: 46.87 -> 35.51\textpm \ 0.25 | Helped |
| Add CGKD (ablation) | DA with StyleGAN-v2 · 100-shot Obama | FID: 46.87 -> 36.18 \textpm \ 0.22 | Helped |
| Add CGKD (ablation) | KD-DLGAN without CGKD · CIFAR-10, 10% data | LPIP: 0.204 -> 0.221 | Helped |
| Add CGKD (ablation) | KD-DLGAN without CGKD · CIFAR-100, 10% data | LPIP: 0.237 -> 0.264 | Helped |
| Use CLIP instead of TCL or BLIP as teacher (ablation) | DA with KD-DLGAN · CIFAR-10, 10% data | FID: 14.98 \textpm \ 0.09; 15.74 \textpm \ 0.10; 14.20 \textpm \ 0.06 | No clear effect |
Closest papers by meaning
- KDGAN: Knowledge Distillation with Generative Adversarial Networks (NeurIPS 2018 · similarity 0.90)
- Wavelet Knowledge Distillation: Towards Efficient Image-to-Image Translation (CVPR 2022 · similarity 0.90)
- Distilling Portable Generative Adversarial Networks for Image Translation (AAAI 2020 · similarity 0.89)
- Taming the Tail in Class-Conditional GANs: Knowledge Sharing via Unconditional Training at Lower Resolutions (CVPR 2024 · similarity 0.88)
- MCGAN: Enhancing GAN Training with Regression-Based Generator Loss (AAAI 2025 · similarity 0.87)
- Small Scale Data-Free Knowledge Distillation (CVPR 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.