The network topology used in this work was an Elman network, which are multi-layer back-propagation networks with a feedback connection from the output of the hidden layer to its input. This feedback path allows Elman networks to learn to recognize and generate temporal patterns, as well as spatial patterns[7].