Bagging ensemble of SVM based on negative correlation learning

A new support vector machine (SVM) ensemble algorithm based on negative correlation learning is studied in this paper. This approach can produce individual SVMs whose errors tend to be negatively correlated, so the diversity is emphasized among individual SVMs in an ensemble. This method is applied in modeling of leaching process of hydrometallurgy. The empirical results show that the method does consistently improve the prediction accuracy versus basic bagging algorithms and single SVM algorithms for leaching process.