Artificial Neural Networks in Rule-based Decision Systems

While building neural models in decision methods based on logic functions and relevant expert knowledge the proper network structure may be chosen according to the given information of the problem solved. There has been presented a certain approach to designing one-directional artificial neural networks with special structure created to solve a specific decision problem based on neural models of logic functions. The input-output relations of the rules in knowledge base are presented in the matrix form. There has been designed a general algorithm for modification of connections between neurons.