Pacific Institute for the Mathematical Sciences - PIMS

Anomaly detection in communications networks via hypergraphs

Anomaly detection is the early recognition of significant events that
occur within a communications network. The use of graph metrics such as
graph edit distance and graph diameters to model changes in networks
and to detect anomaly has been well documented. Possible extensions to
metrics on hypergraphs have been proposed as a method of encapsulating
more information into the network. In this talk, we will describe the
various graph metrics studied, propose ways to extend them into
hypergraphs and provide some initial empirical results using these
hypergraph methods. This work is part of my honours thesis at the
University of Queensland with Diane Donovan and Elizabeth
Billington.

4:00pm-5:00pm, WMAX 216