Dresden 2014 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
SOE: Fachverband Physik sozio-ökonomischer Systeme
SOE 10: Networks, From Topology to Dynamics I (joint with DY and BP)
SOE 10.4: Talk
Tuesday, April 1, 2014, 15:45–16:00, GÖR 226
Automatic discovery of plausible network models — Telmo Menezes1,2 and •Camille Roth1 — 1Centre Marc Bloch Berlin, CNRS — 2Centre d'Analyse et de Mathématique Sociales, CNRS/EHESS
A methodology is proposed to discover plausible network generators for complex networks. Generators are defined as computer programs that define local morphogenetic behaviors. We employ a machine learning technique inspired by biological Darwinism to look for generators that produce synthetic networks which match a number of metrics on target real networks. We use a number of metrics that capture both global and fine-grained structural characteristics of networks. Remarkably, when applied on networks stemming from prototypical models of the Erdös-Rényi or Barabási-Albert sort, our approach generally discovers the exact original generator. Empirical validation of our methodology is then presented in the form of a number of plausible generators for a series of five real networks, including a simple brain and a social network.