Skip to Main Content
Currently, combining multiple neural networks appears to be a very promising approach in improving neural network generalisation since it is very difficult, if not impossible, to develop a perfect single neural network. In this paper, individual networks are developed from bootstrap re-samples of the original training and testing data sets. Instead of combining all the developed networks, this paper proposes selective combination techniques: forward selection. These techniques essentially combine those individual networks that, when combined, can significantly improve model generalisation. The proposed techniques are applied to modelling irreversible exothermic reaction in CSTR. Application results demonstrate that the proposed techniques can significantly improve model generalisation and perform better than aggregating all the individual networks.
Date of Conference: 6-8 June 2006