Committee-based sample selection for probabilistic classifiers

In many real-world learning tasks it is expensive to acquire a sufficient number of labeled examples for training. This paper investigates methods for reducing annotation cost by sample selection. ...