Reproduce “Learning to Defer with Limited Expert Predictions” 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
Patrick Hemmer, Lukas Thede, Michael Vössing, Johannes Jakubik, Niklas Kühl
The paper presents a method for training AI–radiologist teams to decide when to defer chest X-ray airspace-opacity judgments with few expert predictions. Using ground-truth labels, it learns an embedding and expertise predictor that estimates whether the radiologist is correct, then generates artificial predictions for remaining cases. On NIH data, six predictions per class achieved 99.12% of reference system accuracy for one radiologist and finding in a limited experiment.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.