Research map / Label-efficient and robust learning / Domain adaptation and generalization
Semi-Supervised Domain Adaptation via Minimax Entropy
Kuniaki Saito, Donghyun Kim, Stan Sclaroff, Trevor Darrell, Kate Saenko
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 |
|---|---|---|---|
| add entropy maximization on the classifier (minimax) instead of plain entropy minimization (ablation) | ENT (ResNet34) · DomainNet mean 1-shot | accuracy: 62.6 -> 66.4 | Improved |
| add entropy maximization on the classifier (minimax) instead of plain entropy minimization (ablation) | ENT (AlexNet) · DomainNet mean 1-shot | accuracy: 29.1 -> 44.2 | Improved |
| use the l2-normalized cosine classifier with temperature instead of a standard linear layer (ablation) | MME (AlexNet) · DomainNet R to C 1-shot | accuracy: 44.9 -> 48.9 | Improved |
| use the l2-normalized cosine classifier with temperature instead of a standard linear layer (ablation) | S+T (AlexNet) · DomainNet R to C 1-shot | accuracy: 41.4 -> 43.3 | Improved |
| use separate BN statistics for labeled and unlabeled batches (ablation) | ENT (ResNet34) · DomainNet | accuracy: 63.6 -> 68.9 | Improved |
| use separate BN statistics for labeled and unlabeled batches (ablation) | MME (ResNet34) · DomainNet | accuracy: 69.5 -> 69.6 | No clear effect |
| add virtual adversarial training loss on unlabeled target | S+T (AlexNet) · DomainNet R to C 3-shot | accuracy: 47.1 -> 46.1 | Hurt |
| add MME to S+T | S+T (ResNet34) · DomainNet mean 1-shot | accuracy: 56.9 -> 66.4 | Improved |
| add entropy minimization (ENT) to S+T | S+T (AlexNet) · DomainNet mean 1-shot | accuracy: 40.0 -> 29.1 | Hurt |
| add DANN domain classifier to S+T | S+T (VGG16) · DomainNet mean 1-shot | accuracy: 50.5 -> 45.4 | Hurt |
| add MME in unsupervised DA setting | Source-only (AlexNet) · DomainNet mean | accuracy: 37.2 -> 40.2 | Improved |
| add entropy minimization (ENT) to S+T with more labels | S+T · DomainNet | accuracy: 5 labeled examples per class | Mixed |
Closest papers by meaning
- Attract, Perturb, and Explore: Learning a Feature Alignment Network for Semi-supervised Domain Adaptation (ECCV 2020 · similarity 0.91)
- A Balanced and Uncertainty-aware Approach for Partial Domain Adaptation (ECCV 2020 · similarity 0.90)
- Minimal-Entropy Correlation Alignment for Unsupervised Deep Domain Adaptation (ICLR 2018 · similarity 0.90)
- Forget More to Learn More: Domain-specific Feature Unlearning for Semi-supervised and Unsupervised Domain Adaptation (ECCV 2024 · similarity 0.90)
- Bridging Theory and Algorithm for Domain Adaptation (ICML 2019 · similarity 0.89)
- Few-Shot Adversarial Domain Adaptation (NeurIPS 2017 · 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.
