Research map / Label-efficient and robust learning / Domain adaptation and generalization
MineGAN: Effective Knowledge Transfer From GANs to Target Domains With Few Images
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 mining before generator finetuning (ablation) | BigGAN transfer to Places365 · ImageNet → Places365 | FID: 82.3 -> 78.4 | Helped |
| Add mining before generator finetuning (ablation) | BigGAN transfer to ImageNet target classes · ImageNet → ImageNet | FID: 61.8 -> 52.3 | Helped |
| Add mining before generator finetuning (ablation) | Multi-source Progressive GAN transfer · Red vehicles (car:bus 0.3:0.7) | FID: 67.3 -> 61.2 | Helped |
| Add mining before generator finetuning (ablation) | Multi-source Progressive GAN transfer · Tower | FID: 69.2 -> 62.4 | Helped |
| Add mining before generator finetuning (ablation) | Multi-source Progressive GAN transfer · Bedroom | FID: 58.9 -> 54.7 | Helped |
| Replace max-based selection/backpropagation with mean over generated images (ablation) | MineGAN selector for Cars and Buses · Red vehicles (car:bus 0.3:0.7) | Selector probability assigned to Bus: 0.49 -> 0.66 | Hurt |
| Increase the number of fully connected miner layers (ablation) | BigGAN MineGAN · Places365 Arch | FID | Helped |
| Increase the miner from three to four fully connected layers (ablation) | BigGAN MineGAN · Places365 Arch | FID | No clear effect |
| Use conditional rather than unconditional mining (ablation) | MineGAN on MNIST · MNIST average, on-manifold | Classifier error: 5.7 -> 2.2 | Helped |
| Use conditional rather than unconditional mining (ablation) | MineGAN on MNIST · MNIST average, off-manifold | Classifier error: 6.0 -> 3.2 | Helped |
Closest papers by meaning
- Efficient Conditional GAN Transfer With Knowledge Propagation Across Classes (CVPR 2021 · similarity 0.89)
- Training Language GANs from Scratch (NeurIPS 2019 · similarity 0.88)
- DeGAN: Data-Enriching GAN for Retrieving Representative Samples from a Trained Classifier (AAAI 2020 · similarity 0.87)
- Re-GAN: Data-Efficient GANs Training via Architectural Reconfiguration (CVPR 2023 · similarity 0.87)
- Taming the Tail in Class-Conditional GANs: Knowledge Sharing via Unconditional Training at Lower Resolutions (CVPR 2024 · similarity 0.87)
- MCGAN: Enhancing GAN Training with Regression-Based Generator Loss (AAAI 2025 · 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.