@misc{indiciae1dec266f79c5,
  title = {Similarity-as-Evidence: Calibrating Overconfident VLMs for Interpretable and Label-Efficient Medical Active Learning},
  author = {Zhuofan Xie and Zishan Lin and Jinliang Lin and Jie Qi and Shaohua Hong and Shuo Li},
  year = {2026},
  url = {https://arxiv.org/abs/2602.18867},
  note = {Source identifier: 2602.18867}
}