An experimental comparison of ensemble learning methods on decision boundaries

This paper presents an experimental comparison on different kinds of neural network ensemble learning methods on a patter classification problems. To summarize, there are three ways of designing neural network ensembles in these methods: independent training, sequential training and simultaneous training. The purpose of such comparison is not only to illustrate the learning behavior of different neural network ensemble learning methods, but also to cast light on how to design more effective neural network ensembles. The experimental results show that the decision boundary of the trained neural network ensemble by negative correlation learning is almost as good as the optimum decision boundary.