Neural network architecture of information systems

Abstract

The problem of process automation for the development of information systems based on the application of the original neural network architecture is considered. An analysis of existing approaches to the automation of information systems design is carried out. Recommendations for the information systems architecture, aimed at reducing the negative impact of human factor, are formulated. A general concept of neural network architecture in the form of a structural model is presented. Definitions of the main entities and components are given. The key differences of the neural network architecture are: The independence of the key entities of information systems and the possibility of automation of their design and interaction based on the use of neural networks; isolation of the mathematical software of architecture; separation of models of information processes and functional elements from control systems and information representation systems; taking into account the influence of the environment on the processes of movement of information flows, the elements of control and system visualization; the possibility of adapting structural units of information systems to the characteristics of the subject area, the parameters of user equipment without the need to make significant changes to the architecture. The concept of a neural network data channel, its structure and generalized mathematical software are considered. The decomposition of the structural mode is implemented. The structural diagrams of each entity of the neural network architecture of information systems, the description of the main components, the neural network data channels used to connect the entities and their components are presented. The scope of application of the neural network architecture is analyzed.