Learning Large-Scale Patterns in Complex Network Data

  • Date
    January 7, 2014
  • Time
    10:30 AM
  • Location
    366 WVH

Abstract

Networks provide a rich and mathematically principled approach to characterizing the structure of complex social, biological and even technological systems. A common step in understanding the structure and function of some real-world network is to first characterize its large-scale structure, which may be modeled as a probability distribution over connections at scales proportional to the size of the entire network, and which may depend on rich meta-data associated with the vertices or edges. This “generative model” approach allows us to leverage powerful techniques from both machine learning and statistical physics to make data-driven statements about observed, latent, past or future connections.

In this talk, I will present a unified probabilistic framework for modeling large-scale network structure. Special cases include inferring latent “block” structure in weighted and unweighted networks and inferring ultra-metric or hierarchical patterns, among others. Along the way, I will describe some of the general methodological difficulties posed by large-scale inference in networks, and highlight selected scientific applications from computational biology. To close, I will describe some of my current projects in this area, including modeling non-stationarity and detecting anomalies in evolving networks.

As time allows, I will briefly summarize some of my work in other areas of computational research, which includes pattern discovery in the global statistics of terrorism and war, modeling and forecasting rare events, large-scale patterns in ecology and evolutionary biology, predictability in team competitions and other sports, and social dynamics in online games.

 

Brief Biography

Aaron Clauset is an Assistant Professor in the Department of Computer Science and the BioFrontiers Institute at the University of Colorado Boulder, and is External Faculty at the Santa Fe Institute. He received a PhD in Computer Science, with distinction, from the University of New Mexico, a BS in Physics, with honors, from Haverford College, and was an Omidyar Fellow at the prestigious Santa Fe Institute.

He is an internationally recognized expert on network science, data science, and machine learning for complex systems. His work has appeared in prestigious scientific venues like Nature, Science, PNAS, JACM, ICML, STOC, SIAM Review, and Physical Review Letters, and has been covered in the popular press by the Wall Street Journal, The Economist, Discover Magazine, New Scientist, Wired, Miller-McCune, the Boston Globe and The Guardian.

Host: Alan Mislove