On the Function Approximation Using Neural Networks

ROMANIAN JOURNAL OF INFORMATION SCIENCE AND TECHNOLOGY
Volume 1, Number 1, 1998, 55 - 64

On the Function Approximation Using Neural Networks
Ion CIUCA
Research Institute for Informatics
Str. General Averescu 8-10, 71316 Bucuresti, Romania
E-mail: ciucai@u3.ici.ro

Abstract.
The paper deals with the approximation of continuous functions by feedforward neural networks. It presents an explicit formula for function approximation implementable as a four-layer feedforward neural network using sigmoidal and bell-shaped activation functions. These four-layer feedforward neural networks have the same number of neurons in the hidden layers as the four-layer neural networks constructed by Ito, [16], and Cardaliaguet-Euvrard, [2].