home · mobile · calendar · colloquia · 2008-2009 · 

Colloquium - Clauset

Inferring Large-Scale Structural Patterns in Complex Networks
Santa Fe Institute

Networks have become a powerful tool for studying complex systems: they provide an abstraction of a system's interacting parts that is both general enough to encompass important features of real systems and simple enough to offer clear insights and general results. Already, networks have become a central tool in understanding a wide range of biological, social, and technological phenomena.

Until recently, most work on networks focused on simple statistical regularities, such as degree distributions and correlations, centrality measures, etc. These measures do yield some insight, but they capture only a fraction of the complexity of real-world networks. Increasingly, progress on important questions of structure and function depend on going beyond these measures to understand the origins and functional significance of large-scale structural patterns, such as modules and hierarchies.

In this talk, I'll briefly describe my work developing algorithms to automatically infer these large-scale patterns. Using the example of hierarchical structure, I'll describe a flexible generative model and a principled technique for inferring it directly from network data. A hierarchical organization, it turns out, can simultaneously explain many of the simple statistical regularities mentioned above, as well as generalize a single network to an ensemble of statistically similar networks, and make highly accurate predictions about missing links.

As time allows, I'll briefly mention my work in other areas, e.g., on the evolution of species body sizes.

Hosted by Debra Goldberg and Andrew Martin.

Department of Computer Science
University of Colorado Boulder
Boulder, CO 80309-0430 USA
May 5, 2012 (14:13)