Research map / Label-efficient and robust learning / Domain adaptation and generalization
Adversarial Multiple Source 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 |
|---|---|---|---|
| soft (log-sum-exp) instead of hard max over source domains (ablation) | MDAN · Amazon B+E+K/D | accuracy: 0.7797 -> 0.8065 | Helped |
| soft instead of hard objective (ablation) | MDAN · Digits T+S+D/M | accuracy: 0.663 -> 0.687 | Helped |
| soft instead of hard objective (ablation) | MDAN · WebCamT camera A, 5 sources | MAE: 1.4021 -> 1.1942 | Helped |
| multisource adversarial adaptation (MDAN) vs combined-source DANN | MLP · Amazon B+E+K/D | accuracy: 0.7886 -> 0.8065 | Helped |
| MDAN vs combined-source DANN | MLP · Amazon B+D+E/K | accuracy: 0.8639 -> 0.8626 | No clear effect |
| MDAN vs best single-source DANN | digit CNN · Digits M+T+D/S | accuracy: 0.818 -> 0.816 | No clear effect |
| combine all sources into one domain for DANN instead of source-only | DANN · Digits S+M+D/T | accuracy: 0.938 -> 0.925 | Hurt |
| MDAN vs DANN | FCN · WebCamT camera A, 5 sources | MAE: 1.4156 -> 1.1942 | Helped |
| add more (more distant) source cameras (ablation) | MDAN · WebCamT | MAE | Mixed |
Closest papers by meaning
- Multi-Adversarial Domain Adaptation (AAAI 2018 · similarity 0.92)
- Adversarial Discriminative Domain Adaptation (CVPR 2017 · similarity 0.91)
- Bridging Theory and Algorithm for Domain Adaptation (ICML 2019 · similarity 0.90)
- Transferable Adversarial Training: A General Approach to Adapting Deep Classifiers (ICML 2019 · similarity 0.90)
- f-Domain Adversarial Learning: Theory and Algorithms (ICML 2021 · similarity 0.90)
- A DIRT-T Approach to Unsupervised Domain Adaptation (ICLR 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.