next up previous contents
Next: Example PEP: A ``straight-line'' Up: Training architecture Previous: Raw training data

Parametrised event primitives

Parametrised event primitives (PEPs) are a way to incorporate background knowledge into the architecture. They define different ``shapes'' in the channels we are examining. A PEP consists of the following:

Each parametrised event primitive (PEP) is described by a set of parameters along with a function for extracting instances of that event primitive. Thus, by applying the finding function, we can find all the events that appear to be an instance of an event primitive we are considering. Example PEPs are discussed below.





Mohammed Waleed Kadous
Tue Oct 6 13:04:40 EST 1998