Research map / Label-efficient and robust learning / Domain adaptation and generalization
Generate to Adapt: Aligning Domains Using Generative Adversarial Networks
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 Generate to Adapt instead of source-only training | SVHN→MNIST · SVHN→MNIST | accuracy: 60.3 ± 1.5 -> 92.4 ± 0.9 | Helped |
| Use Generate to Adapt rather than the strongest reported baseline | Office domain adaptation · Office average | accuracy: 84.3 -> 86.5 | Helped |
| Use Generate to Adapt instead of source-only training | CAD→PASCAL with VGG16 · CAD→PASCAL | accuracy: 38.1 ± 0.4 -> 50.4 ± 0.6 | Helped |
| Adapt the source-trained model using Generate to Adapt | VISDA-C ResNet-18 · VISDA-C validation | accuracy: 35.3 -> 63.1 | Helped |
| Adapt the source-trained model using Generate to Adapt | VISDA-C ResNet-50 · VISDA-C validation | accuracy: 40.2 -> 69.5 | Helped |
| Adapt the source-trained model using Generate to Adapt | VISDA-C ResNet-152 · VISDA-C validation | accuracy: 44.5 -> 77.1 | Helped |
| Add target-data adversarial training to source-only training (ablation) | Office A→W · Office A→W | accuracy: 68.4 -> 80.5 | Helped |
| Add the auxiliary classifier C2 to the real/fake discriminator (ablation) | Generate to Adapt · Office A→W | accuracy: 80.5 -> 89.5 | Helped |
| Vary the generator noise dimension from 32 to 512 (ablation) | Generate to Adapt · SVHN→MNIST | accuracy | Mixed |
Closest papers by meaning
- Unsupervised Pixel-Level Domain Adaptation With Generative Adversarial Networks (CVPR 2017 · similarity 0.92)
- Adversarial Discriminative Domain Adaptation (CVPR 2017 · similarity 0.92)
- Learning From Synthetic Data: Addressing Domain Shift for Semantic Segmentation (CVPR 2018 · similarity 0.91)
- Duplex Generative Adversarial Network for Unsupervised Domain Adaptation (CVPR 2018 · similarity 0.90)
- Adversarial Feature Augmentation for Unsupervised Domain Adaptation (CVPR 2018 · similarity 0.90)
- Few-Shot Adversarial Domain Adaptation (NeurIPS 2017 · similarity 0.90)
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.