Artificial Neural Networks as a Tool for Economic Development Planning

Abstract

Mathematical models that simulate the networks of nerve cells of living organisms are actively used for forecasting purposes, but they have the potential in terms of planning. Artificial neural networks may be of interest as a platform for managing economic processes. The article examines evolution of economic development planning, explores the possibilities of using artificial neural networks for planning purposes from economic cybernetics point of view. The author relies on the definition of a plan as an algorithm of actions coordinated in time and space, capable of transforming the system. Neural networks were compared with the basic laws of cybernetics in the article, different types of neural networks were demarcated and the purposes of their use were analyzed. It is concluded that feedforward artificial neural networks can only be used for forecasting purposes, since they give a statistical average weighted result, and not exact calculations. Recurrent neural networks can be used for “preplanning” or indicative planning. It is possible to take individual indicators and consider them in dynamics, but this model is fraught with disparities and distortions in production. The author concludes that indicative planning is no longer relevant in complex systems and market diversity. It is shown that the direct and feedback algorithm and the computing apparatus are necessary for directive and strategic planning of the economic development, this is possible only in the case of the synthesis of neural networks and the dynamic model of interbranch-intersectoral balance developed by N.I. Veduta.