This paper shows how to construct a generative model for graph structure. We commence from a sample of graphs where the correspondences between nodes are unknown ab initio. We also work with graphs where there may be structural differences present, i.e. variations in the number of nodes in each graph and the edge-structure. The idea underpinning the method is to embed the nodes of the graphs into a vector space by performing kernel PCA on the heat kernel. The co-ordinates of the nodes are determined by the eigenvalues and eigenvectors of the Laplacian matrix, together with a time parameter which can be used to scale the embedding. Node correspondences are located by applying Scott and Longuet-Higgins algorithm to the embedded nodes. We capture variations in graph structure using the covariance matrix for corresponding embedded point-positions. We construct a point distribution model for the embedded node positions using the eigenvalues and eigenvectors of the covariance matrix. We show how to use this model to both project individual graphs into the eigenspace of the point-position covariance matrix and how to fit the model to potentially noisy graphs to reconstruct the Laplacian matrix. We illustrate the utility of the resulting method for shape-analysis using data from the COIL database.


Adjacency Matrix Heat Kernel Graph Structure Laplacian Matrix Sample Graph 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


  1. 1.
    Christmas, W.J., Kittler, J., Petrou, M.: Structural matching in computer vision using probabilistic relaxation. IEEE Transactions on Pattern Analysis and Machine Intelligence 17(8), 749–764 (1995)CrossRefGoogle Scholar
  2. 2.
    Wilson, R.C., Hancock, E.R.: Structural matching by discrete relaxation. IEEE Transactions on Pattern Analysis and Machine Intelligence 19(6), 634–648 (1997)CrossRefGoogle Scholar
  3. 3.
    Bagdanov, A.D., Worring, M.: First Order Gaussian Graphs for Efficient Structure Classification. Pattern Recognition 36, 1311–1324 (2003)MATHCrossRefGoogle Scholar
  4. 4.
    Wong, A.K.C., Constant, J., You, M.L.: Random Graphs Syntactic and Structural Pattern Recognition. World Scientific, Singapore (1990)Google Scholar
  5. 5.
    Heckerman, D., Geiger, D., Chickering, D.M.: Learning Bayesian Networks: The combination of knowledge and statistical data. Machine Learning 20, 197–243 (1995)MATHGoogle Scholar
  6. 6.
    Scott, G.L., Longuett-Higgins, H.C.: An algorithm for associating the features of two images. Proceedings of the Royal Society of London 244, 21–26 (1991)CrossRefGoogle Scholar
  7. 7.
    Lozano, M.A., Escolano, F.: ACM Attributed Graph Clustering for Learning Classes of Images. In: Hancock, E.R., Vento, M. (eds.) GbRPR 2003. LNCS, vol. 2726, pp. 247–258. Springer, Heidelberg (2003)CrossRefGoogle Scholar
  8. 8.
    Bunke, H., et al.: Graph Clustering Using the Weighted Minimum Common Supergraph. In: Hancock, E.R., Vento, M. (eds.) GbRPR 2003. LNCS, vol. 2726, pp. 235–246. Springer, Heidelberg (2003)CrossRefGoogle Scholar
  9. 9.
    Jain, B.J., Wysotzki, F.: Central Clustering of Attributed Graphs. Machine Learning 56, 169–207 (2004)MATHCrossRefGoogle Scholar
  10. 10.
    Scott, G.L., Longuet-Higgins, H.C.: An Algorithm for Associating the Features of two Images. Proceedings of the Royal Society of London 244, 21–26 (1991)CrossRefGoogle Scholar
  11. 11.
    Harris, C.G., Stephens, M.J.: A Combined Corner and Edge Detector. In: Fourth Alvey Vision Conference, pp. 147–151 (1994)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2006

Authors and Affiliations

  • Bai Xiao
    • 1
  • Edwin R Hancock
    • 1
  1. 1.Department of Computer ScienceUniversity of YorkYorkUK

Personalised recommendations