University of Southern California

ISI Site Signature

Aram Galstyan
 Complex Adaptive Networks  
  Our society is becoming increasingly dependent on various technological networks, such as transportation systems, electrical power distribution grids, computer networks, and so on. As those networks evolve and grow in complexity, their dynamical behavior is becoming difficult to understand and predict. Part of our research focused on developing computational models of growth and evolution of such complex adaptive networks.  
 Statistical Mechanics of Inference and Learning  
  Many problems in machine learning can be framed as reconstructing a process based on noisy observations of that process. For instance, in object tracking one is interested in inferring object's trajectory based on noisy observations of its spatial location. In social network analysis, on is interested in inferring the underlying social structure based on the pattern of interactions between individuals. Despite apparent differences, those problems share some common underlying theoretical issues. One such issue is whether there are fundamental limits on one's ability to infer such processes, and how to quantify those limits.