Návrh simulátoru založeného na umělých neuronových sítích pro identifikaci vstupně výstupních funkcí systémů

Abstract

This thesis is focuses on proposal of the simulator of the systems based on artificial neural networks and its use in static forecasting of economic time series with comparing the results of statistical methods. Theoretical part defines statistical methodology and principles of artificial neural networks. Thesis thereafter describe proposal of the simulator for RBF and Perceptron architecture and its use in static forecast of economic time series and compares results with ARIMA models.