A hierarchical Bayesian scheme for nonlinear dynamical system reconstruction and prediction with neural nets

A hierarchical Bayesian scheme with neural nets is used to reconstruct nonlinear dynamical systems. Typical examples include chaotic time series prediction and energy demand prediction of a building. The latter class of problems helps in saving energy and reduction of CO/sub 2/ emissions. A difference between these two classes of problems lies in the fact that the former gives rise to autonomous dynamical systems while the latter leads to non-autonomous dynamical systems.