A computational scheme for Reasoning in Dynamic Probabilistic Networks

A computational scheme for reasoning about dynamic systems using (causal) probabilistic networks is presented. The scheme is based on the framework of Lauritzen and Spiegel­ halter {1988), and may be viewed as a gen­ eralization of the inference methods of clas­ sical time-series analysis in the sense that it allows description of non-linear, multi­ variate dynamic systems with complex con­ ditional independence structures. Further , the scheme provides a met hod for efficient backward smoothing and possibilities for effi­ cient, approximate forecasting methods. The scheme has been implemented on top of the HUGIN shell.