AnchorAL is a computationally efficient active learning method that addresses class imbalance by dynamically selecting a small, balanced subpool of instances to run the active learning strategy on, promoting the discovery of minority instances.
