Knowledge-enhanced Parameter-efficient Transfer Learning with METER for medical vision-language tasks.

Journal: Journal of biomedical informatics
Published Date: May 8, 2025

Abstract

OBJECTIVE: The full fine-tuning paradigm becomes impractical when applying pre-trained models to downstream tasks due to significant computational and storage costs. Parameter-efficient fine-tuning (PEFT) methods can alleviate the issue. However, solely applying PEFT methods leads to sub-optimal performance owing to the domain gap between pre-trained models and medical downstream tasks.

Authors

  • Xudong Liang
    School of Computer Engineering and Science, Shanghai University, Shanghai, China.
  • Jiang Xie
    Soil and Fertilizer & Resources and Environmental Institute, Jiangxi Academy of Agricultural Sciences, Nanchang, China.
  • Jinzhu Wei
    School of Medicine, Shanghai University, Shanghai 200444, China.
  • Mengfei Zhang
    Computer Science Department, University of Southern California, Los Angeles, CA, U.S.A.
  • Haoyang Zhang
    School of Computer Engineering and Science, Shanghai University, Shanghai 200444, China.

Keywords

External Resources

Popular Topics
Recent Journals