Label-efficient Learning

We develop label-efficient learning methods—including semi-supervised and weakly supervised learning—to achieve high-accuracy medical image analysis with limited annotated data, reducing the burden of expert annotation in clinical settings.