Constructing an ensemble Learning Model by using Euclidean distance

The support vector machine (SVM) has a good generalization performance, but the classification result of the SVM in some real problems is often unsatisfied. Because SVM is sensitive to the noisy data and it may not be effective under the high level of noise. To improve the performance of SVM in the noisy environment, we propose an ensemble learning model to address the noise problem in this work. First, we employ the noise-tolerant probabilistic Support Vector Machine. Then a Naive Bayesian classifier is established in the model. Finally the decision of the two classifiers is appropriately combined to give the final decision. We use the Euclidean distance to complete the integration based on a probabilistic interpretation. The ensemble learning model is evaluated on an artificial dataset for a classification task. Compared with single classifier, the ensemble learning model exhibits good performance in the noisy environment.