Conventional artificial neural networks are based on greatly simplified models of biological neurons. In particular only one
path exists to carry an input signal to the neuron body. The design of a parallel path artificial micro-net is described which
can be used as the basic building block to construct networks. Preliminary training results are presented.