Evolving multiple neural networks

In this paper we apply genetic algorithms to the automatic generation of neural networks as well as the biological inspiration of neural networks to successfully adapt to environments. The network produced by this method can be customized for a special objective because the network is selected by the objective function. The final goal in designing a classifier is to achieve the best performance for a given classification. It has been observed that some methods of combining networks consistently outperform a single network. Therefore, we also investigate the performance of combining multiple evolving neural networks. Financial and medical data are used to test the network's performance.