Abstract
It is pointed out that from a point of view of parallel processing, neural network models belong to the more general class of communicating sequential processes. Sequential operating neurons communicate through input and output connections with other neurons working in parallel. While such a viewpoint brings nothing novel to the theory of neural models as such, it can be used to make model development, simulations, model alterations, comparisons, and testing easier. This is because the theory of communicating sequential processes has an efficient implementation in the Occam language and transputer processors. The basic neural network model specification is given, together with code for model implementation. As an example, Kohonen's self-organizing map model is used.

This publication has 2 references indexed in Scilit: