A novel approach for determining the optimal number of hidden layer neurons for FNN’s and its application in data mining
Shuxiang Xu, Ling Chen · eCite Digital Repository (University of Tasmania) · 2008
Optimizing the number of hidden layerneurons for an FNN (feedforward neural network) tosolve a practical problem remains one of the unsolvedtasks in this research area. In this paper we reviewseveral mechanisms in the neural networks literaturewhich have been used for determining an optimalnumber of hidden layer neuron (given an application),propose our new approach based on some mathematicalevidence, and apply it in financial data mining.Compared with the existing methods, our new approachis proven (with mathematical justification), and can beeasily handled by users from all application fields.