A class+1 sigmoidal activation functions for FFANNs

A limited number of activation functions have been utilized in practice for feedforward artificial neural networks. A class of sigmoidal functions is defined and another function, which is the envelope of the derivatives of the members of the defined class, is shown to be sigmoidal. The functions defined are shown to satisfy the requirements of the universal approximation theorem(s). (C) 2003 Published by Elsevier Science B.V.