Confidence-calibrated covariate shift correction for few-shot classification in vision-language models
Confidence calibration and covariate shift correction for few-shot classification in vision-language models.
CVPR Workshops, 2025.
Note
Corrects for the distribution shift between a vision-language model’s pretraining data and a new target domain, and calibrates the confidence it reports in the few-shot regime.
Reference
Confidence-calibrated covariate shift correction for few-shot classification in vision-language models CVPR Workshops, 2025. https://scholar.google.com/scholar?q=%22Confidence-calibrated%20covariate%20shift%20correction%20for%20few-shot%20classification%20in%20vision-language%20models%22
The publisher record holds the author list.
Details
- TypeConference paper
- AreasFoundation models for biology and medicine
More from the index
- PreviousMulti-Modal LLMs in Agriculture: A Comprehensive ReviewIEEE Transactions on Automation Science and Engineering
- NextA Layered Self-Supervised Knowledge Distillation Framework for Efficient Multimodal Learning on the EdgearXiv