TITLE: Inference for unlabelled graphs

SPEAKER: Professor Peter Bickel

ABSTRACT:

A great deal of attention has recently been paid to determining sub-communities on the basis of relations, corresponding to edges, between individuals, corresponding to vertices   of an unlabelled graph (Newman, SIAM Review 2003; Airoldi et al JMLR 2008; Leskovec &  Kleinberg et al SIGKDD 2005). We have developed a nonparametric framework for probabilistic ergodic models of infinite unlabelled graphs (PNAS2009)and made some connections with modularities arising in the physics literature   and community models in the social sciences.A fundamental difficulty in implementing these procedures is  computational complexity.We  develop   approaches  which  bypass these difficulties.

 (This is joint work with Aiyou Chen and Liza Levina)