Research map / Label-efficient and robust learning / Test-time adaptation
Multi-Label Test-Time Adaptation with Bayesian Conditional Priors
QiRu Li, Ao Zhou, Zhiwei Jiang, Zifeng Cheng, Cong Wang, Yafeng Yin, Qing Gu
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 2 anchors instead of 1 (average PMI over anchors) (ablation) | BCP ViT-B/16 · Average of 5 datasets | mAP: 71.79 -> 69.93 | Hurt |
| use 3 anchors instead of 1 (average PMI over anchors) (ablation) | BCP ViT-B/16 · Average of 5 datasets | mAP: 71.79 -> 67.94 | Hurt |
| add warmup that delays correction until statistics stabilize (ablation) | BCP ViT-B/16 · Average of 5 datasets | mAP: 71.79 -> 71.95 | Improved |
| apply BCP prior correction to frozen CLIP on Anchor-Hit samples | CLIP ViT-B/16 · Anchor-Hit subset average | mAP: 72.73 -> 80.33 | Improved |
| apply BCP prior correction to frozen CLIP on Anchor-Miss samples | CLIP ViT-B/16 · Anchor-Miss subset | mAP: +1.48 on COCO2014, -3.75 on VOC2007 | Mixed |
| use BCP instead of BEM | CLIP RN50 · COCO2014 | adaptation time and mAP: 0.24s -> 0.112s; 51.58 -> 61.69 | Improved |
| use BCP instead of SCA | CLIP ViT-B/16 · Average of 5 datasets | mAP: 69.24 -> 71.79 | Improved |
| use BCP instead of DOTA | CLIP RN50 · NUSWIDE | mAP: 38.88 -> 50.34 | Improved |
| use BCP instead of SCA on tail labels | CLIP ViT-B/16 · COCO14 tail split | mean AP: 67.61 -> 69.22 | Improved |
| apply BCP to CLIP with many labels per image | CLIP RN50 · COCO2014 labels >= 8 | mAP: 41.07 -> 51.64 | Improved |
| make test stream more non-i.i.d. (gamma 0.1 to 0.001) (ablation) | BCP ViT-B/16 · Average of 5 datasets | mAP: 69.15 -> 68.66 | No clear effect |
| shuffle test stream order with five seeds (ablation) | BCP ViT-B/16 · COCO14 | mAP: 65.48, 65.40, 65.48, 65.49, 65.42 | No clear effect |
Closest papers by meaning
- Multi-Label Test-Time Adaptation with Bound Entropy Minimization (ICLR 2025 · similarity 0.89)
- Bayesian Test-Time Adaptation for Vision-Language Models (CVPR 2025 · similarity 0.89)
- Efficient Test-Time Adaptation of Vision-Language Models (CVPR 2024 · similarity 0.88)
- Noisy Test-Time Adaptation in Vision-Language Models (ICLR 2025 · similarity 0.88)
- Efficient Test-time Adaptation through Candidate Verification and Divergence Shifts (NeurIPS 2026 · similarity 0.87)
- Efficient and Context-Aware Label Propagation for Zero-/Few-Shot Training-Free Adaptation of Vision-Language Model (ICLR 2025 · similarity 0.87)
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.
