Workshop 3: Advanced computational intelligence techniques for identification, control and optimization of nonlinear systems

Neural networks and fuzzy systems are natural candidates as approximators of a nonlinear time series or dynamical system, due to their intrinsic nonlinearity and computational simplicity. Under the stationarity hypothesis for the system generating the data, the NARX (Nonlinear Auto-Regressive with an eXogenous (X) variable) neural networks are able to solve the nonlinear identification problem. The multilayer feedforward and recurrent neural networks types are employed.