Multiplicative-additive neural networks with active neurons

An artificial neural network is a flexible mathematical structure which is capable of identifying complex nonlinear relationships between input and output data sets. Such neural networks have been characterized by passive neurons that are not able to select and estimate their own inputs. In a new approach, which corresponds in a better way to the actions of human nervous system, the connections between several neurons are not fixed but change in dependence on the neurons themselves. This paper deals with the applications of the self-organization multiplicative-additive algorithm with active neurons to prediction models of river flow. The nonlinear multiplicative-additive model approach is shown to provide better representation of the weekend average water inflow forecasting in comparison to the models based on the Box-Jenkins method, currently in use on the Brazilian Electrical Sector