Components of Active Learning Sample Clauses

Components of Active Learning. Similar to standard passive learning methods, in active learning we start with an initial training set ( 0, 0), where e { } e Y { } X { } 0 = x1,. .., xN are the N feature vectors with d di- mensions, i.e. xi d, and 0 = y1,. .., yN are the corresponding class labels, i.e. yi 1,. .., C . In this paper, we assume the number of classes C as given, but usually larger than 2. We note that, in contrast to passive learning, active learners usually can deal with much smaller initial