Research system using neural-type algorithms for decision making

The authors propose a system consisting of a set of subsystems allowing one to automate the procedure of decision making. Signal transformation at a large number of subsystems is realized by similar models of neural networks. The implementation of various required operations (signal transformations) is carried out by controlling the form of the coupling function between neuron-type elements; the character of subelement activation; and adaptive architecture of a system signal or data flow, depending on varying ways between neural network subsystems. The system is constructed according to an open-type, allowing one to use the knowledge of experts at every stage of the processing.<>