D-Confidence: An Active Learning Strategy which Efficiently Identifies Small Classes
Nuno Escudeiro, Alípio Jorge · 2010
In some classification tasks, such as those related to the automatic building and maintenance of text corpora, it is expensive to obtain labeled examples to train a classifier. In such circumstances it is common to have massive corpora where a few examples are labeled (typically a minority) while others are not. Semi-supervised learning techniques try to leverage the intrinsic information in unlabeled examples to improve classification models. However, these techniques assume that the labeled examples cover all the classes to learn which might not stand. In the presence of an imbalanced class distribution getting labeled examples from minority classes might be very costly if queries are randomly selected. Active learning allows asking an oracle to label new examples, that are criteriously selected, and does not assume a previous knowledge of all classes. D-Confidence is an active learning approach that is effective when in presence of imbalanced training sets. In this paper we discuss the performance of d-Confidence over text corpora. We show empirically that d-Confidence reduces the number of queries required to identify examples from all classes to learn when compared to confidence, a common active learning criterion.