| Abstrakt: |
This paper presents the results of a series of research trials of the developed recurrent neural networks (RNN) architecture xMANN (external memory memorya-augmented neural network) and a successful experience of integration of a recurrent neural network into a process information system. The effectiveness of process chain analysis and control depends on the potential ability to control the associative-spatial interaction of signals in recurrent neural networks. Mathematical modelling allows to formalise many problems and ways to solve them. However, there is a possibility of limitations of the mathematical model, which also needs to be taken into account in the simulation. We present a step-by-step integration plan to effectively create an information system with advanced information analysis and processing capabilities, as the use of a neural network allows you to work with partially structured tasks, while it is possible to implement the function of generating alternative solutions. [ABSTRACT FROM AUTHOR] |
