Multivariable Function Approximation With Multilayer Perceptron Neural Networks:- Obasi, Nwanne E

ABSTRACT viii In this research work we have studied and analysed a multidimensional function approximation with the Multilayer Perceptron (MLP) neural network built into the Statistical Package for Social Science (SPSS). Our results have shown that the MLP neural network performs well as a function approximator for both single variable and multivariable functions. This fact is guaranteed by the function approximation theorem for neural networks. We remark that it is the function approximation capabilities ofthe Artificial Neural Network (ANN) that makes them so powerful and applicable in the solution of differential equaf'-'nc