Oral Candidacy - Jian Xu
|Start:||4/19/2016 at 9:30AM|
|End:||4/19/2016 at 12:30PM|
|Location:||384 Nieuwland Hall|
Faculty and students are welcome to attend the presentation portion of the defense.
April 19, 2016
384 Nieuwland Science Hall (iCeNSA)
Adviser: Dr. Nitesh Chawla
Dr. David Lodge Dr. Tijana Milenkovic Dr. Zoltan Toroczkai
"Representing Data as Networks: New Methods and Insights"
Network science is a powerful tool for studying complex systems, yet to ensure the correctness of network analysis results, the network (as the input) has to be a sufficiently accurate representation of the underlying data. It begets the question that how should one represent data as networks, without losing important patterns in the raw data. When representing sequential data from complex systems such as global shipping traffic or clickstream traffic as networks, conventional network representations that implicitly assume the Markov property (first-order dependency) can quickly become limiting. This assumption holds that when movements are simulated on the network, the next movement depends only on the current node, discounting the fact that the movement may depend on several previous steps. Work in this proposal focuses on the development of novel methods to extract higher-order dependencies from data and embed them into a network. This enriched network representation can then provide new insights in the downstream network analyses, such as clustering, ranking, and anomaly detection. The said methods are then leveraged to solve practical problems such as the evaluation of species invasion risks driven by the global shipping network. The proposed network representation as a general method will have impacts on a broad range of real-world problems that have been studied through the lens of network science.