Given a set of graphs, the median graph is defined as the graph which has the smallest sum of distances (SOD) to all the graphs in the set. It has been proposed as a tool to obtain the representative of such a set. In spite of its potential applications, the existing algorithms are computationally complex and have a very limited applicability. In this paper we propose a new approach for the exact computation of the median graph based on graph embedding in vector spaces. Graphs are embedded into a vector space and the median is computed in the vector domain. After that, the median graph is recovered from this median vector. Experiments on a synthetic database show that our approach outperforms the previous existing exact algorithms both on computation time and number of SOD computations.


  1. 1.
    Jiang, X., Münger, A., Bunke, H.: On median graphs: Properties, algorithms, and applications. IEEE Trans. Pattern Anal. Mach. Intell. 23(10), 1144–1151 (2001)CrossRefGoogle Scholar
  2. 2.
    Münger, A.: Synthesis of prototype graphs from sample graphs. Diploma Thesis, University of Bern (in German) (1998)Google Scholar
  3. 3.
    Ferrer, M.: Theory and algorithms on the median graph. application to graph-based classification and clustering. PhD Thesis, Universitat Autònoma de Barcelona (2007)Google Scholar
  4. 4.
    Hlaoui, A., Wang, S.: Median graph computation for graph clustering. Soft Comput. 10(1), 47–53 (2006)CrossRefGoogle Scholar
  5. 5.
    White, D., Wilson, R.C.: Mixing spectral representations of graphs. In: 18th International Conference on Pattern Recognition (ICPR 2006), Hong Kong, China, August 20-24, pp. 140–144. IEEE Computer Society, Los Alamitos (2006)CrossRefGoogle Scholar
  6. 6.
    Bunke, H.: On a relation between graph edit distance and maximum common subgraph. Pattern Recognition Letters 18(8), 689–694 (1997)CrossRefGoogle Scholar
  7. 7.
    Sanfeliu, A., Fu, K.: A distance measure between attributed relational graphs for pattern recognition. IEEE Transactions on Systems, Man and Cybernetics 13(3), 353–362 (1983)CrossRefzbMATHGoogle Scholar
  8. 8.
    Bunke, H., Allerman, G.: Inexact graph matching for structural pattern recognition. Pattern Recognition Letters 1(4), 245–253 (1983)CrossRefzbMATHGoogle Scholar
  9. 9.
    Schenker, A., Bunke, H., Last, M., Kandel, A.: Graph-Theoretic Techniques for Web Content Mining (Machine Perception and Artificial Intelligence) (Series in Machine Perception and Artificial Intelligence). World Scientific Publishing Co., River Edge (2005)zbMATHGoogle Scholar
  10. 10.
    Riesen, K., Neuhaus, M., Bunke, H.: Graph embedding in vector spaces by means of prototype selection. In: Escolano, F., Vento, M. (eds.) GbRPR 2007. LNCS, vol. 4538, pp. 383–393. Springer, Heidelberg (2007)CrossRefGoogle Scholar
  11. 11.
    Krause, E.F.: Taxicab Geometry: An Adventure in Non-Euclidean Geometry. Dover, New York (1986)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2008

Authors and Affiliations

  • Miquel Ferrer
    • 1
  • Ernest Valveny
    • 1
  • Francesc Serratosa
    • 2
  • Horst Bunke
    • 3
  1. 1.Centre de Visió per Computador, Dep. Ciències de la ComputacióUniversitat Autònoma de BarcelonaBellaterraSpain
  2. 2.Departament d’Enginyeria Informàtica i MatemàtiquesUniversitat Rovira i VirgiliTarragonaSpain
  3. 3.Department of Computer Science and Applied MathematicsUniversity of BernBernSwitzerland

Personalised recommendations