Reproduce “Evaluating Federated Vision-Language and Vision-Only Models for Privacy-Aware Breast Cancer Imaging” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Malithi Mithsara Wanniarachchi Kankanamge, Ning Yang, Sree Bash Debnath, Md Zobaer Islam, Abdur R. Shahid
The authors evaluate federated vision-language models for two breast-imaging tasks, keeping images local and sharing updates. They compare privacy noise and CLIP without its text pathway. On BCD, acquisition year reaches ROC-AUC 0.974, close to CLIP’s 0.979. Removing text preserves performance and reduces communication; thermography results remain uncertain. Neither result establishes clinical accuracy or which image cues models use.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.