Multiobjective learning of complex recurrent neural network

Modeling of complex systems requires specific methods of model assessment. Local assessment of the model of complex system reduces to independent evaluations of quality of its parts. However, such a procedure does not necessarily lead to an optimal model of the whole complex system. To do this, global assessment should be used because it takes into account connections between elements. It is also possible to combine these two approaches. In this work, we make use of multiobjective approach to assess quality of model of complex system. Complex neural network is used as a model of dynamic complex system. As an example of complex dynamic system for experimental purposes, chemical process is considered.