Dynamical Operators and Clustering for Multiplex Data

Dr. Daryl Deford (host Nussinov/Dev), Dartmouth College
October 19, 2017 at 3:00 pm
241 Compton
Event Description 

In recent years, multilayer analyses of network data have become increasingly common and extending techniques originally designed for single--layer networks has generated a significant amount of interest. Two main families of models have been proposed for describing multiplex data, one emphasizing structural features and another motivated by dynamical processes. In this talk we will give a broad description of the properties of both models and discuss the related background and open questions, motivated by applications to specific data sets. The setting of graph decomposition provides a clear example distinguishing the behavior of the models. Using a series of three numerical experiments we will demonstrate the advantages of dynamical models for spectral clustering and community detection problems.

Coffee: 2:45 pm, 241 Compton