Bidirectional Uncertainty-Based Active Learning for Open Set Annotation

arXiv:2402.15198v1 Announce Type: new
Abstract: Active learning (AL) in open set scenarios presents a novel challenge of identifying the most valuable examples in an unlabeled data pool that comprises data from both known and unknown classes. Traditional methods prioritize selecting informative examples with low confidence, with the risk of mistakenly selecting unknown-class examples with similarly low confidence. Recent methods favor the most probable known-class examples, with the risk of picking simple already mastered examples. In this paper, we attempt to …

abstract active learning annotation arxiv challenge class confidence cs.lg data examples low novel pool risk set type uncertainty