Analysis of an evolutionary RBFN design algorithm, CO2RBFN, for imbalanced data sets
Pattern Recognition Letters
| Hi-index | 0.00 |
In this paper we take our effort to achieve a fast and accurate classifier: a BVB (BAM-Vote Box)-based framework is presented for text categorization by using ensemble method. The central idea is that combining two-class classifications for the multi-class tasks. This framework generates associating terms and extending the set of basis element, and includes a feature selection method, which can reduce the number of features and thus reduce computation and improve accuracy. A greedy algorithm is used in feature selection to overcome a serious bias of the unbalanced feature set. Then the new classifier, a BVB-based composite model, is introduced. Experimental results show that our approach can obtain high quality classifications and far less time-consuming than other neural networks, such as back propagation (BP) network and radial basis function (RBF) network, etc.