Research map / Label-efficient and robust learning / Domain adaptation and generalization
Understanding Self-Training for Gradual 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 |
|---|---|---|---|
| Sequentially self-train across gradual domains instead of self-training on target data only | Self-training · Gaussian | accuracy: 49.6$\pm$0.0 -> 98.8$\pm$0.0 | Helped |
| Sequentially self-train across gradual domains instead of pooling unlabeled data | Self-training · Gaussian | accuracy: 92.5$\pm$0.1 -> 98.8$\pm$0.0 | Helped |
| Sequentially self-train across gradual domains instead of self-training on target data only | Self-training · Rotating MNIST | accuracy: 33.0$\pm$2.2 -> 87.9$\pm$1.2 | Helped |
| Sequentially self-train across gradual domains instead of pooling unlabeled data | Self-training · Rotating MNIST | accuracy: 38.0$\pm$1.6 -> 87.9$\pm$1.2 | Helped |
| Sequentially self-train across gradual domains instead of self-training on target data only | Self-training · Portraits | accuracy: 76.9$\pm$2.1 -> 83.8$\pm$0.8 | Helped |
| Sequentially self-train across gradual domains instead of pooling unlabeled data | Self-training · Portraits | accuracy: 78.9$\pm$3.0 -> 83.8$\pm$0.8 | Helped |
| Use hard pseudolabels instead of probabilistic labels (ablation) | Gradual ST · Gaussian | accuracy: 90.5$\pm$1.9 -> 99.3$\pm$0.0 | Helped |
| Use hard pseudolabels instead of probabilistic labels (ablation) | Gradual ST · Rotating MNIST | accuracy: 44.1$\pm$2.3 -> 83.8$\pm$2.5 | Helped |
| Use hard pseudolabels instead of probabilistic labels (ablation) | Gradual ST · Portraits | accuracy: 80.1$\pm$1.8 -> 82.6$\pm$0.8 | Helped |
| Add explicit regularization during self-training (ablation) | Gradual ST · Gaussian | accuracy: 84.6$\pm$1.1 -> 99.3$\pm$0.0 | Helped |
| Add explicit regularization during self-training (ablation) | Gradual ST · Rotating MNIST | accuracy: 45.8$\pm$2.5 -> 83.8$\pm$2.5 | Helped |
| Add explicit regularization during self-training (ablation) | Gradual ST · Portraits | accuracy: 76.5$\pm$1.0 -> 82.6$\pm$0.8 | Helped |
Closest papers by meaning
- Understanding Gradual Domain Adaptation: Improved Analysis, Optimal Path and Beyond (ICML 2022 · similarity 0.91)
- Self-Training with Dynamic Weighting for Robust Gradual Domain Adaptation (NeurIPS 2025 · similarity 0.91)
- Gradual Domain Adaptation via Manifold-Constrained Distributionally Robust Optimization (NeurIPS 2024 · similarity 0.88)
- Enhancing Evolving Domain Generalization through Dynamic Latent Representations (AAAI 2024 · similarity 0.87)
- A General Representation-Based Approach to Multi-Source Domain Adaptation (ICML 2025 · similarity 0.86)
- Pseudo-Labeling for Unsupervised Domain Adaptation with Kernel GLMs (NeurIPS 2026 · similarity 0.86)
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.