The Bayesian formalism for combining multiple decision of a neural network ensemble

A new methodology for improving the performance and training of neural network classifiers is presented. The main idea is based on using redundant classifiers in an ensemble in order to guarantee the best generalisation ability of the classifier. As compared to previous designs, a novel method for output combination based on weighted averaging is introduced. The proposed technique consist in considering the classes independently of one another and calculating the importance parameters, i.e. the weights, for individual outputs of the networks. In order to draw a comparison with previous methods, a real data medical benchmark is used.