Nonlinear system identification using spatiotemporal neural networks

Abstract

The so-called spatiotemporal neural network is considered. This is a neural network where the conventional weight multiplication operation is replaced by a linear filtering operation. A training algorithm is derived for such networks. The problem of nonlinear system identification is considered as an application for spatiotemporal networks. Nonlinear system identification is one of the problems in the systems area, with limited success for results based on conventional methods. Neural network approaches are encouraging, but further exploration is needed. The capability of the spatiotemporal neural networks to identify nonlinear systems is explored through a simple example using the derived learning rule. The simulation results are encouraging, though testing of the identification method on a real-world system is still under investigation.< >