A Bayesian network approach to explaining time series with changing structure

摘要: Many examples exist of multivariate time series where dependencies between variables change over time. If these changing are not taken into account, any model that is learnt from the data will average different dependency structures. Paradigms try to explain underlying processes and observed events in must explicitly changes order allow non-experts analyse understand such data. In this paper we have developed a method for generating explanations takes account structure. We make use dynamic Bayesian network with hidden nodes. introduce representation search technique learning models test it on synthetic real-world an oil refinery, both which contain compare our existing EM-based Results very promising include sample explanations, generated refinery dataset.