Research map / Label-efficient and robust learning / Test-time adaptation
SODA: Robust Training of Test-Time Data Adaptors
Zige Wang, Yonggang Zhang, Zhen Fang, Long Lan, Wenjing Yang, Bo Han
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 |
|---|---|---|---|
| Use SODA instead of leaving the model unadapted | ResNet-50 · CIFAR-10-C | accuracy (%): 72.39 -> 82.55 | Improved |
| Use SODA instead of leaving the model unadapted | ResNet-50 · CIFAR-100-C | accuracy (%): 41.41 -> 52.41 | Improved |
| Use SODA instead of leaving the model unadapted | ResNet-50 · ImageNet-C | accuracy (%): 31.36 -> 42.14 | Improved |
| Use first-order gradients rather than ZOO, with SODA-R training settings | SODA · CIFAR-10-C | accuracy (%): 82.55 -> 88.39 | Improved |
| Increase data-adaptor depth from 0 to 3 ResNet blocks (ablation) | SODA · CIFAR-10-C | accuracy (%): 82.55 -> 78.65 | Hurt |
| Increase data-adaptor depth from 0 to 3 ResNet blocks (ablation) | SODA-R · CIFAR-10-C | accuracy (%): 86.07 -> 88.51 | Improved |
| Add perturbation regularization (ablation) | SODA · CIFAR-10-C | accuracy (%): 82.55 -> 73.40 | Hurt |
| Add perturbation regularization (ablation) | SODA-R · CIFAR-100-C | accuracy (%): 58.11 -> 60.31 | Improved |
| Use Adam instead of SGD (ablation) | SODA · CIFAR-100-C | accuracy (%): 52.41 -> 49.75 | Hurt |
| Use Adam instead of SGD (ablation) | SODA-R · CIFAR-100-C | accuracy (%): 58.32 -> 60.31 | Improved |
| Add dropout layers (ablation) | SODA · CIFAR-100-C | accuracy (%): 43.96 -> 5.47 | Hurt |
| Add dropout layers (ablation) | SODA-R · CIFAR-100-C | accuracy (%): 58.27 -> 60.31 | Improved |
Closest papers by meaning
- Distribution Alignment for Fully Test-Time Adaptation with Dynamic Online Data Streams (ECCV 2024 · similarity 0.88)
- Towards Real-World Test-Time Adaptation: Tri-net Self-Training with Balanced Normalization (AAAI 2024 · similarity 0.86)
- A Layer Selection Approach to Test Time Adaptation (AAAI 2025 · similarity 0.86)
- Test-time Adaptation in Non-stationary Environments via Adaptive Representation Alignment (NeurIPS 2024 · similarity 0.85)
- CONTRAST: Continual Multi-source Adaptation to Dynamic Distributions (NeurIPS 2024 · similarity 0.85)
- Protected Test-Time Adaptation via Online Entropy Matching: A Betting Approach (NeurIPS 2024 · similarity 0.85)
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.
