Research map / Label-efficient and robust learning / Test-time adaptation
Parameter-Free Online Test-Time Adaptation
Check 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 |
|---|---|---|---|
| Apply LAME output correction instead of leaving predictions unadapted | ImageNet-pretrained models · 7 test scenarios, Original RN-50 | average accuracy: 52.07 -> 55.70 | Helped |
| Use LAME output correction rather than the non-adaptive baseline | ImageNet-pretrained models · 7 test scenarios, all architectures | average accuracy: 56.9 -> 60.7 | Helped |
| Use LAME output correction rather than network adaptation | ImageNet-pretrained models · Non-i.i.d. prior-shift scenarios | accuracy: 6.7% | Helped |
| Use LAME output correction rather than network adaptation | ImageNet-pretrained models · ImageNet-V2 with non-i.i.d. prior shift | accuracy: 15% | Helped |
| Apply LAME in standard i.i.d. likelihood shifts | ImageNet-pretrained models · i.i.d. likelihood shifts | accuracy: 0.5 % | Hurt |
| Replace max-pooling with average-pooling when mapping source classes to target superclasses (ablation) | LAME · Superclass class mapping | accuracy | Helped |
| Use kNN, linear, or RBF affinity kernels (ablation) | LAME · 7 test scenarios across 5 architectures | average accuracy | No clear effect |
| Vary batch size across a wide range (ablation) | LAME · 7 test scenarios | average accuracy improvement over baseline: close-to-4% | No clear effect |
Closest papers by meaning
- Free on the Fly: Enhancing Flexibility in Test-Time Adaptation with Online EM (CVPR 2025 · similarity 0.90)
- Realistic Test-Time Adaptation of Vision-Language Models (CVPR 2025 · similarity 0.90)
- Efficient Test-Time Adaptation of Vision-Language Models (CVPR 2024 · similarity 0.89)
- The Illusion of Progress? A Critical Look at Test-Time Adaptation for Vision-Language Models (NeurIPS 2025 · similarity 0.89)
- DOTA: Distributional Test-time Adaptation of Vision-Language Models (NeurIPS 2025 · similarity 0.89)
- Frustratingly Easy Test-Time Adaptation of Vision-Language Models (NeurIPS 2024 · similarity 0.88)
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.