active learning is a machine learning paradigm that aims to improve the
performance of a model by strategically selecting and querying unlabeled data.
One effective selection strategy is to base it on the model's predictive
uncertainty, which can be interpreted as a measure of how info