Research map / Vision-language and multimodal / CLIP and zero-shot
Prototype-Based Test-Time Adaptation of Vision-Language Models
Zhaohong Huang, Yuxin Zhang, Wenjing Liu, Fei Chao, Rongrong Ji
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 adaptive decay rate beta instead of raw confidence scores for prototype updates (ablation) | PTA · ImageNet | accuracy (%): 0.10 -> 70.28 | Improved |
| use adaptive decay rate beta instead of raw confidence scores (ablation) | PTA · EuroSAT | accuracy (%): 11.11 -> 61.57 | Improved |
| add the interpolation weight w (0 -> 0.01) between visual and text-anchored prototypes (ablation) | PTA · ImageNet | accuracy (%): 0.10 -> 70.28 | Improved |
| increase w from 0.01 to 0.1 (ablation) | PTA · ImageNet | accuracy (%): 70.28 -> 68.00 | Hurt |
| increase w from 0.0001 to 0.001 to 0.01 (ablation) | PTA · Cross-Domain | accuracy (%): 69.35 -> 69.35 -> 69.38 | No clear effect |
| add PTA to GS-Bias | GS-Bias · Cross-dataset (11 incl. ImageNet) | accuracy (%): 67.35 -> 69.61 | Improved |
| add PTA to MTA | MTA · Cross-dataset (11 incl. ImageNet) | accuracy (%): 66.33 -> 69.33 | Improved |
| add PTA to TDA | TDA · Cross-dataset (11 incl. ImageNet) | accuracy (%): 68.07 -> 69.58 | Improved |
| add PTA to TDA on EuroSAT | TDA · EuroSAT | accuracy (%): 62.51 -> 61.88 | Hurt |
| place confidently misclassified samples first in the stream (poison size All vs Normal) (ablation) | PTA · ImageNet | accuracy (%): 70.3 -> 70.3 | No clear effect |
| use PTA instead of TDA under low non-iid streams (gamma=0.1) | TDA · non-iid average, 11 datasets | accuracy (%): 67.7 -> 67.8 | No clear effect |
| use PTA instead of TDA for speed | TDA · ImageNet efficiency | FPS: 44.6 -> 81.8 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- ADAPT — ADAPT: Vision-Language Navigation With Modality-Aligned Action Prompts (CVPR 2022) · better on 2 of 3 shared results
- GS-Bias — GS-Bias: Global-Spatial Bias Learner for Single-Image Test-Time Adaptation of Vision-Language Models (ICML 2025) · better on 1 of 1 shared result
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
- The Illusion of Progress? A Critical Look at Test-Time Adaptation for Vision-Language Models (NeurIPS 2025 · similarity 0.95)
- Efficient Test-Time Adaptation of Vision-Language Models (CVPR 2024 · similarity 0.94)
- DOTA: Distributional Test-time Adaptation of Vision-Language Models (NeurIPS 2025 · similarity 0.93)
- Efficient Test-time Adaptation through Candidate Verification and Divergence Shifts (NeurIPS 2026 · similarity 0.93)
- Realistic Test-Time Adaptation of Vision-Language Models (CVPR 2025 · similarity 0.92)
- BoostAdapter: Improving Vision-Language Test-Time Adaptation via Regional Bootstrapping (NeurIPS 2024 · similarity 0.92)
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.
