direkt zum Inhalt springen

direkt zum Hauptnavigationsmenü

Sie sind hier

TU Berlin

Page Content

New Directions in Graph Clustering

Sharon Bruckner (Freie Universität Berlin)
Graph clustering is the task of grouping the nodes of a graph into clusters, taking into consideration the edge structure of the graph in such a way that there should be many edges within each cluster and relatively few between the clusters. Under this somewhat ambiguous umbrella one finds a variety of interpretations and approaches, with contributions from physicists and statisticians, biologists and computer scientists. This talk will first survey some recent trends in this field, highlighting popular algorithms and estimating the areas where further solutions are still needed. In the second part of the talk I will describe the approach of the Biocomputing group to graph clustering and present work done with Christof Schütte, Tim Conrad, Nataša Djurdjevac and Marco Sarich. This approach is based on continuous random walks on the graph, identifying clusters with "metastable sets" where the random walker stays for extended periods of time.

18.05.2011 14:15
Sharon Bruckner
FR 6510

Back to overview.