SoQal: Selective Oracle Questioning in Active Learning
This work addresses the challenge of underutilized unlabeled healthcare data by making active learning more efficient, though it is incremental as it builds on existing active learning methods.
The paper tackled the problem of reducing the cost and time of active learning in healthcare by proposing SoQal, a strategy that dynamically decides when to request labels from an oracle, resulting in up to a 35% reduction in label requests while maintaining competitive performance under label noise.
Large sets of unlabelled data within the healthcare domain remain underutilized. Active learning offers a way to exploit these datasets by iteratively requesting an oracle (e.g. medical professional) to label instances. This process, which can be costly and time-consuming is overly-dependent upon an oracle. To alleviate this burden, we propose SoQal, a questioning strategy that dynamically determines when a label should be requested from an oracle. We perform experiments on five publically-available datasets and illustrate SoQal's superiority relative to baseline approaches, including its ability to reduce oracle label requests by up to 35%. SoQal also performs competitively in the presence of label noise: a scenario that simulates clinicians' uncertain diagnoses when faced with difficult classification tasks.