Neural networks approach to discovering temporal correlations

Many real-life problems require discovering temporal correlations between behavior of a complex object and relatively rare events initiated by this behavior, or correlating with it. It is supposed that an event is preceded by some combination of the values describing the object in a known range of delays. In this paper, a neural network-based method for analyzing such objects is proposed with the view of revealing the morphological and dynamic features that initiate an event or precede its occurrence.