Introduction
Artificial neural networks (ANNs, see [Hay94, Kos92]) are models of the brain’s cognitive process. In contrast with conventional single-processor computers, the brain has a multiprocessor architecture that is highly inter- connected. This architecture has been described as parallel distributed processing (PDP). PDP has many advantages over single-processor models for many difficult computer science problems. It is one of the reasons that the brain, with its slow, error-prone individual hardware units (neurons) so much outperforms even supercomputers in some areas.