Research map / Image generation / GANs
Effective Data Augmentation with Multi-Domain Learning GANs
Shin'ya Yamaguchi, Sekitoshi Kanai, Takeharu Eda
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 |
|---|---|---|---|
| Train the GAN jointly on target and outer domains instead of fine-tuning from the outer domain | GAN-based data augmentation · CIFAR-100 | Top-1 accuracy (%): 25.9±0.5 -> 28.9±0.5 | Improved |
| Train the GAN jointly on target and outer domains instead of fine-tuning from the outer domain | GAN-based data augmentation · FGVC-Aircraft | Top-1 accuracy (%): 24.1±0.4 -> 27.3±0.9 | Improved |
| Train the GAN jointly on target and outer domains instead of fine-tuning from the outer domain | GAN-based data augmentation · Indoor Scene Recognition | Top-1 accuracy (%): 24.0±0.2 -> 26.1±0.7 | Improved |
| Add conventional data augmentation to DF | DF · CIFAR-100 | Top-1 accuracy (%): 30.7±0.7 -> 32.1±0.7 | Improved |
| Add conventional data augmentation to DF | DF · FGVC-Aircraft | Top-1 accuracy (%): 29.6±0.9 -> 31.2±0.7 | Improved |
| Add conventional data augmentation to DF | DF · Indoor Scene Recognition | Top-1 accuracy (%): 31.0±0.3 -> 32.4±1.7 | Improved |
| Add discriminator rejection sampling (DRS) (ablation) | DF · CIFAR-100 | Top-1 accuracy (%): 28.3±0.7 -> 28.9±0.5 | Improved |
| Add discriminator rejection sampling (DRS) (ablation) | DF · FGVC-Aircraft | Top-1 accuracy (%): 27.0±0.5 -> 27.3±0.9 | Improved |
| Add discriminator rejection sampling (DRS) (ablation) | DF · Indoor Scene Recognition | Top-1 accuracy (%): 25.4±0.1 -> 26.1±0.7 | Improved |
| Use the relevance-and-diversity outer-dataset metric M and avoid the worst outer dataset (ablation) | DF · CIFAR-100 | Top-1 accuracy (%): 25.5±0.3 -> 28.9±0.5 | Improved |
| Use the relevance-and-diversity outer-dataset metric M and avoid the worst outer dataset (ablation) | DF · FGVC-Aircraft | Top-1 accuracy (%): 24.2±0.3 -> 27.3±0.9 | Improved |
| Use the relevance-and-diversity outer-dataset metric M and avoid the worst outer dataset (ablation) | DF · Indoor Scene Recognition | Top-1 accuracy (%): 24.2±0.3 -> 26.1±0.7 | Improved |
Closest papers by meaning
- Adversarial Feature Augmentation for Unsupervised Domain Adaptation (CVPR 2018 · similarity 0.88)
- Augmentation-Aware Self-Supervision for Data-Efficient GAN Training (NeurIPS 2023 · similarity 0.86)
- Differentiable Augmentation for Data-Efficient GAN Training (NeurIPS 2020 · similarity 0.86)
- Generate to Adapt: Aligning Domains Using Generative Adversarial Networks (CVPR 2018 · similarity 0.86)
- RadialGAN: Leveraging multiple datasets to improve target-specific predictive models using Generative Adversarial Networks (ICML 2018 · similarity 0.85)
- Adversarial Discriminative Domain Adaptation (CVPR 2017 · similarity 0.85)
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.
