Research map / Label-efficient and robust learning / Domain adaptation and generalization
Transferable Adversarial Training: A General Approach to Adapting Deep Classifiers
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 |
|---|---|---|---|
| Adversarial training with transferable examples (TAT) instead of adversarial feature adaptation | ResNet-50 classifier · Office-31 | accuracy (avg): 88.4 vs 87.5 (CDAN), 76.1 (ResNet-50) | Helped |
| Adversarial generation of transferable examples from category classifier gradient (removed) (ablation) | TAT · Office-31 | accuracy (avg): 88.4 full vs 85.8 without | Helped |
| Adversarial generation from domain discriminator gradient (removed) (ablation) | TAT · Office-31 | accuracy (avg): 88.4 full vs 86.8 without | Helped |
| TAT on ImageCLEF-DA gains are small over strongest baseline | TAT · ImageCLEF-DA | accuracy (avg): 88.9 vs 87.1 (CDAN) | Helped |
| TAT adaptation on Office-Home | ResNet-50 · Office-Home | accuracy (avg): 65.8 vs 63.8 (CDAN) | Helped |
| TAT on VisDA-2017 synthetic-to-real | ResNet-50 · VisDA-2017 | accuracy: 71.9 vs 70.0 (CDAN) | Helped |
| TAT applied on mSDA representations | mSDA features · Multi-Domain Sentiment | accuracy (avg): 86.3 vs 85.0 (TDA) | Helped |
| TAT on BoW input with two-layer classifier | BoW features · Multi-Domain Sentiment | accuracy (avg): 82.4 vs 82.3 (mSDA BoW) | No clear effect |
Closest papers by meaning
- Unsupervised Domain Adaptation with Residual Transfer Networks (NeurIPS 2016 · similarity 0.91)
- A DIRT-T Approach to Unsupervised Domain Adaptation (ICLR 2018 · similarity 0.91)
- Adversarial Discriminative Domain Adaptation (CVPR 2017 · similarity 0.90)
- Transferability vs. Discriminability: Batch Spectral Penalization for Adversarial Domain Adaptation (ICML 2019 · similarity 0.90)
- Cooperative and Adversarial Learning: Co-enhancing Discriminability and Transferability in Domain Adaptation (AAAI 2023 · similarity 0.90)
- Adversarial Multiple Source Domain Adaptation (NeurIPS 2018 · 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.