XAL: EXplainable Active Learning Makes Classifiers Better Low-resource Learners

arXiv:2310.05502v2 Announce Type: replace
Abstract: Active learning (AL), which aims to construct an effective training set by iteratively curating the most formative unlabeled data for annotation, has been widely used in low-resource tasks. Most active learning techniques in classification rely on the model's uncertainty or disagreement to choose unlabeled data, suffering from the problem of over-confidence in superficial patterns and a lack of exploration. Inspired by the cognitive processes in which humans deduce and predict through causal information, we take …

abstract active learning annotation arxiv classification classifiers construct cs.cl data low set tasks training type uncertainty