Active learning with multiple classifiers for multimedia indexing

We propose and evaluate in this paper a combination of Active Learning and Multiple Classifiers approaches for corpus annotation and concept indexing on highly Unbalanced datasets. Experiments were conducted using TRECVID 2008 data and protocol with four different types of video shot descriptors, with two types of classifiers (Logistic Regression and Support Vector Machine with RBF kernel) and with two different active learning strategies (relevance and uncertainty sampling). Results show that the Multiple Classifiers approach significantly increases the effectiveness of the Active Learning. On this dataset, the best performance is reached when 15 to 30% of the corpus is annotated.