Speed-up opportunities for ANN in a time-share parallel environment

Usual optimizations on artificial neural networks (ANN) are mainly algorithmic improvements. Since biological nets make use of massive parallelism, some researchers pursue this direction, which we believe is currently not exploited enough. We discuss such an example of optimization. In this paper, we start by briefly reviewing the frame setting features of a parallel Unix ANN mapping. By slightly increasing the parallelism degree, we have previously obtained some speed-up effects. Still, those effects seemed insignificant, when compared with usual effects obtained by parallelization in an environment based on hardware parallelism. Here we will show with the help of a very simple example problem that the actual effect is much higher, even though the environment on which it is obtained supports only simulated, time-share parallelism.