Research map / Label-efficient and robust learning / Test-time adaptation
Dual Distribution Estimation for Zero-shot Noisy Test-Time Adaptation with VLMs
Wenjie Zhu, Yabin Zhang, Liang Xu, Xin Jin, Wenjun Zeng, Lei Zhang
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 |
|---|---|---|---|
| replace inclusion-only PFDE with PFDE using inclusion and exclusion (ablation) | DDE · ImageNet noisy stream | Acc_H: 72.77 -> 73.52 | Improved |
| add NLDE (ablation) | DDE (no PFDE) · ImageNet noisy stream | Acc_N: 83.96 -> 95.99 | Improved |
| add NLDE to PFDE (in+ex) (ablation) | DDE · ImageNet noisy stream | Acc_H: 73.52 -> 76.82 | Improved |
| add adaptive lambda (ablation) | DDE · ImageNet noisy stream | Acc_H: 76.82 -> 78.00 | Improved |
| increase noise rate from 0.1 to 0.7 (ablation) | DDE · ImageNet noisy stream | Acc_H: 78.00 -> 76.87 | No clear effect |
| scale backbone from RN50 to ViT-L (ablation) | DDE · ImageNet noisy stream | Acc_H: 81.20 (ViT-L), +10.12% over RN50 | Improved |
| increase number of negative labels M from 500 to 1000 (ablation) | DDE · ImageNet noisy stream | Acc_H: 78.00 -> 71.28 | Hurt |
| reduce scaling factor rho from 0.005 to 0.0005 (ablation) | DDE · ImageNet noisy stream | Acc_H: 78.00 -> 77.68 | No clear effect |
| increase exclusion GDA weight beta up to 0.5 (ablation) | DDE · ImageNet noisy stream | Acc_H: 77.42 -> 78.00 | Improved |
| increase cache size Q to 5000 (ablation) | DDE · ImageNet noisy stream | Acc_H | Improved |
| use DDE instead of AdaND | AdaND · ImageNet noisy TTA | Acc_H: 73.09 -> 76.79 | Improved |
| use DDE instead of AdaND for OOD detection | AdaND · ImageNet OOD detection | FPR95: 16.00 -> 9.80 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- AdaNeg — AdaNeg: Adaptive Negative Proxy Guided OOD Detection with Vision-Language Models (NeurIPS 2024) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Noisy Test-Time Adaptation in Vision-Language Models (ICLR 2025 · similarity 0.93)
- Free on the Fly: Enhancing Flexibility in Test-Time Adaptation with Online EM (CVPR 2025 · similarity 0.87)
- Realistic Test-Time Adaptation of Vision-Language Models (CVPR 2025 · similarity 0.87)
- Backpropagation-Free Test-Time Adaptation via Probabilistic Gaussian Alignment (NeurIPS 2025 · similarity 0.87)
- Test-Time Adaptation Induces Stronger Accuracy and Agreement-on-the-Line (NeurIPS 2024 · similarity 0.86)
- Efficient Test-time Adaptation through Candidate Verification and Divergence Shifts (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.
