Speaker:  Dr. Paul Van Dooren 

Abstract:  We present the notion of similarity matrix between two graphs G_{A} and G_{B} as the matrix S whose element S_{ij} measures the similarity of node i in graph G_{A} and node j in graph G_{B}. This matrix can be obtained as a solution of the matrix equation pS=ASB' + A'SB, where A and B are the adjacency matrices of the two graphs. We then show how to construct low rank approximations of this matrix for large scale graphs. This amounts to an optimization algorithm on isometries, and we describe a fixed point algorithm to solve this optimization problem. We give variants of this problem and describe a number of applications. 
Biography:  Paul M. Van Dooren received the engineering degree in computer science and the doctoral degree in applied sciences, both from the Katholieke Universiteit te Leuven, Belgium, in 1974 and 1979, respectively. He held research and teaching positions at the Katholieke Universiteit te Leuven (19741979), the University of Southern California (19781979), Stanford University (19791980), the Australian National University (1984), Philips Research Laboratory Belgium (19801991), the University of Illinois at UrbanaChampaign (19911994), Florida State University (1998) and the Universite Catholique de Louvain (19801991, 1994now) where he is currently a professor of Mathematical Engineering

Presented On:  April 2nd, 2007 
Video:  Click here to see the video 