Multiple models fusion for pattern classification on noise data

An important characteristic of real-world learning process is that the data frequently contains uncertainties. The uncertainties in the datasets deteriorate the learning process. Hence, to properly represent and handle the uncertainty problem is one of the key issues in the decision learning system. This paper offers a multiple models fusion method to address the uncertainty problem, by conducting the fusion of two models, Bayesian classifier and Probabilistic based Noise Aware Support Vector Machine. Specifically, we take the advantage of noise-insensitive characteristic of the Naïve Bayesian classifier, to enhance the noise-tolerant ability of probabilistic information based Support Vector Machine. The method fuses the probabilistic decision information obtained from the two classifiers in a flexible way to give the final decision. Furthermore, the multiple models fusion method is evaluated on an artificial dataset for a classification task. The experiment results show good performance when compared with using only one learning technique in the noise environment.