New construction of Ensemble Classifiers for imbalanced datasets
Yun Zhai, Bingru Yang, Nan Ma, Da Ruan · 2010
Learning in the presence of data imbalances presents a great challenge to machine learning. Imbalanced data sets represent a significant problem because the corresponding classifier has a tendency to ignore samples which have smaller representation in the training sets. In this paper, we propose an ensemble-based learning algorithm as a new ensemble classifier model called as SVM-C5.0 Ensemble Classifier Model, SCECM. SCECM adopts a differentiated sampling rate algorithm (DSRA) based on an improved Adaboost algorithm and further employs unique classifier-selection strategy, novel classifier integration approach and original classification decision-making method. Comparative experimental results show that the proposed approach improves performance for the minority class while preserving the ability to recognize examples from the majority classes.