Milena Mihail - Mathematical models for complex networks

Theoretical quantification as well as experimental evaluation is fundamental in establishing the predictive value of algorithms. In this sense, there is natural overlap between the study of network economics and the modeling of complex networks. We present models of complex networks (first introduced by Bollobas, Janson and Riordan) that are high dimensional generalizations of Erdos Renyi random graphs and capture semantics of correlations of categorical data. We present algorithms to generate such synthetic data in time linear in the number of output nodes and links (and additional polylogarithic factors accounting for data representation and arithmetic.)