TY  - RPRT
TI  - Similarity-as-Evidence: Calibrating Overconfident VLMs for Interpretable and Label-Efficient Medical Active Learning
AU  - Zhuofan Xie
AU  - Zishan Lin
AU  - Jinliang Lin
AU  - Jie Qi
AU  - Shaohua Hong
AU  - Shuo Li
PY  - 2026
UR  - https://arxiv.org/abs/2602.18867
ID  - 2602.18867
ER  -