Research map / Label-efficient and robust learning / Domain adaptation and generalization
Adversarial Feature Augmentation for Unsupervised Domain Adaptation
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 a shared domain-invariant encoder to LS-ADDA (ablation) | LS-ADDA · SVHN→MNIST | accuracy: 0.743 ± 0.028 -> 0.851 ± 0.026 | Helped |
| add feature augmentation to the domain-invariant method (ablation) | Ours (DI) · SVHN→MNIST | accuracy: 0.851 ± 0.026 -> 0.897 ± 0.020 | Helped |
| add a shared domain-invariant encoder to LS-ADDA (ablation) | LS-ADDA · MNIST→USPS P1 | accuracy: 0.914 ± 0.000 -> 0.914 ± 0.000 | No clear effect |
| add feature augmentation to the domain-invariant method (ablation) | Ours (DI) · MNIST→USPS P1 | accuracy: 0.914 ± 0.000 -> 0.923 ± 0.001 | Helped |
| add a shared domain-invariant encoder to LS-ADDA (ablation) | LS-ADDA · MNIST→USPS P2 | accuracy: 0.912 ± 0.003 -> 0.954 ± 0.002 | Helped |
| add feature augmentation to the domain-invariant method (ablation) | Ours (DI) · MNIST→USPS P2 | accuracy: 0.954 ± 0.002 -> 0.962 ± 0.002 | Helped |
| add a shared domain-invariant encoder to LS-ADDA (ablation) | LS-ADDA · USPS→MNIST | accuracy: 0.910 ± 0.004 -> 0.879 ± 0.005 | Hurt |
| add feature augmentation to the domain-invariant method (ablation) | Ours (DI) · USPS→MNIST | accuracy: 0.879 ± 0.005 -> 0.897 ± 0.005 | Helped |
| add a shared domain-invariant encoder to LS-ADDA (ablation) | LS-ADDA · SYN→SVHN | accuracy: 0.908 ± 0.004 -> 0.925 ± 0.002 | Helped |
| add feature augmentation to the domain-invariant method (ablation) | Ours (DI) · SYN→SVHN | accuracy: 0.925 ± 0.002 -> 0.930 ± 0.002 | Helped |
| replace the separate target encoder with a shared domain-invariant encoder (ablation) | LS-ADDA · NYUD RGB→D | accuracy | Helped |
| add feature augmentation to the domain-invariant method (ablation) | Ours (DI) · NYUD RGB→D | accuracy: 0.287 ± 0.002 -> 0.313 ± 0.002 | Helped |
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 0.91)
- Generate to Adapt: Aligning Domains Using Generative Adversarial Networks (CVPR 2018 · similarity 0.90)
- Duplex Generative Adversarial Network for Unsupervised Domain Adaptation (CVPR 2018 · similarity 0.90)
- Maximum Classifier Discrepancy for Unsupervised Domain Adaptation (CVPR 2018 · similarity 0.89)
- A DIRT-T Approach to Unsupervised Domain Adaptation (ICLR 2018 · similarity 0.89)
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.