Abstract
Literature offers several examples of application of neural networks to predict fresh and hardened properties of conventional and self‐compacting concretes. In this paper, these issues are addressed to a relatively recent computational scheme given by the functional networks, by proposing an initial simple application. The algorithm with some appropriate assumptions, reveals a quick convergence and an interesting generalization capability even if the training data set is poor. Copyright © 2009 John Wiley & Sons, Ltd.

This publication has 17 references indexed in Scilit: