NEUROEVOLUTIONAL METHODS FOR DECISION SUPPORT UNDER UNCERTAINTY

The article presents a comparative analysis of the fundamental neuroevolutional methods, which are widely applied for the intellectualization of the decision making support systems under uncertainty. Based on this analysis the new neuroevolutionary method is introduced. It is intended to modify both the topology and the parameters of the neural network, and not to impose additional constraints on the individual. The results of the experimental evaluation of the performance of the methods based on the series of benchmark tasks of adaptive control, classification and restoration of damaged data are carried out. As criteria of the methods evaluation the number of failures and the total number of evolution epochs are used.