Research map / Label-efficient and robust learning / Test-time adaptation
Adaptive Debiasing Tsallis Entropy for Test-Time Adaptation
Xiangyu Wu, Dongming Jiang, Feng Yu, Yueying Tian, Jiaqi Tang, Qing-Guo Chen, Yang Yang, Jianfeng Lu
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 ADTE entropy (vs removing it) (ablation) | ADTE (Templates) · ImageNet variants, ViT-B/16 | accuracy (%): 65.4 -> 66.5 | Improved |
| add ADTE entropy (ablation) | ADTE (Templates) · ImageNet, ViT-B/16 | accuracy (%): 71.1 -> 71.8 | Improved |
| add ADTE entropy (ablation) | ADTE (Templates) · 10 cross-domain datasets, ViT-B/16 | accuracy (%): 66.9 -> 69.0 | Improved |
| add ADTE entropy (ablation) | ADTE (Templates) · ImageNet variants, ViT-L/14 | accuracy (%): 76.3 -> 77.0 | Improved |
| add ADTE entropy (ablation) | ADTE (Templates) · 10 cross-domain datasets, ViT-L/14 | accuracy (%): 73.2 -> 74.8 | Improved |
| add logit adjustment (ablation) | ADTE (Templates) · 10 cross-domain datasets, ViT-B/16 | accuracy (%): 68.8 -> 69.0 | Improved |
| add logit adjustment (ablation) | ADTE (Templates) · ImageNet variants, ViT-B/16 | accuracy (%): 66.1 -> 66.5 | Improved |
| add logit adjustment (ablation) | ADTE (Templates) · 10 cross-domain datasets, ViT-L/14 | accuracy (%): 74.1 -> 74.8 | Improved |
| use ADTE instead of Shannon entropy (SE-LA) (ablation) | ADTE (Templates) · ImageNet, ViT-B/16 | accuracy (%): 71.0 -> 71.8 | Improved |
| use ADTE instead of fixed-q Tsallis entropy (q=0.5) (ablation) | ADTE (Templates) · ImageNet-A, ViT-B/16 | accuracy (%): 64.8 -> 65.5 | Improved |
| increase memory bank size per class from 1 to 40 (ablation) | ADTE (Templates) · ImageNet, ViT-B/16 | accuracy (%): 71.71 -> 71.86 | No clear effect |
| swap Min-Max normalization for Decimal Scaling (ablation) | ADTE (Templates) · ImageNet, ViT-B/16 | accuracy (%): 71.8 -> 72.0 | No clear effect |
Closest papers by meaning
- The Illusion of Progress? A Critical Look at Test-Time Adaptation for Vision-Language Models (NeurIPS 2025 · similarity 0.90)
- Entropy is not Enough for Test-Time Adaptation: From the Perspective of Disentangled Factors (ICLR 2024 · similarity 0.89)
- Efficient Test-Time Adaptation of Vision-Language Models (CVPR 2024 · similarity 0.89)
- Efficient Test-time Adaptation through Candidate Verification and Divergence Shifts (NeurIPS 2026 · similarity 0.89)
- Frustratingly Easy Test-Time Adaptation of Vision-Language Models (NeurIPS 2024 · similarity 0.89)
- BoostAdapter: Improving Vision-Language Test-Time Adaptation via Regional Bootstrapping (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.
