Uniform approximation by a non-convex cone of continuous functions (Q1183170)

The author presents a constructive proof of the result that the set of linear combinations of a bounded sigmoidal function is dense in the space of all continuous functions on the hypercube \(I^ n\). This result has several interesting implications especially in the realizability of a continuous function by a single hidden layer feed forward artificial neural network. Several extensions of their result are also given in this paper.