Forecasting Performance of Security Information for Protected Systems Based on Hybrid Artificial Neural Networks

Abstract

The task of obtaining a scientifically based forecast of indicators and characteristics of information security of protected information systems of the enterprise is very relevant. Currently, the basis of forecasting schemes are formalized methods with a number of significant drawbacks. An alternative to these methods are informal forecasting methods, including those based on soft computing. The analysis of the most common formalized forecasting methods was performed and their unsuitability for solving the problem of obtaining forecasts of information security indicators of an enterprise was justified. Application of hybrid algorithms of intellectual processing is offered as the combination of various approaches provides minimization of shortcomings and maximization of advantages inherent in each of them separately. The paper issues a neurofuzzy network that combines the possibility of using poorly formalized data and training in real time, which can significantly improve the efficiency of forecasting indicators of information security.