PASCAL - Pattern Analysis, Statistical Modelling and Computational Learning

Polytopal Graph Complexity, Matrix Permanents, and Embedding
Francisco Escolano, Edwin Hancock and Miguel Lozano
In: SSPR/SPR 2008, Dec 4-6, 2008, Orlando, USA.


In this paper, we show how to quantify graph complexity in terms of the normalized entropies of convex Birkhoff combinations. We commence by demonstrating how the heat kernel of a graph can be decomposed in terms of Birkhoff polytopes. Drawing on the work of Birkhoff and von Neuman, we next show how to characterise the complexity of the heat kernel. Finally, we provide connections with the permanent of a matrix, and in particular those that are doubly stochastic. We also include graph embedding experiments based on polytopal complexity, mainly in the context of Bioinformatics (like the clustering of protein-protein interaction networks) and image-based planar graphs.

EPrint Type:Conference or Workshop Item (Oral)
Project Keyword:Project Keyword UNSPECIFIED
Subjects:Learning/Statistics & Optimisation
Theory & Algorithms
ID Code:6877
Deposited By:Edwin Hancock
Deposited On:08 April 2010