Relaxation Labelling Using Distributed Neural Networks

  • Jim Austin
Part of the Studies in Computational Intelligence book series (SCI, volume 210)


This chapter presents the Relaxation by Elimination methods (RBE) for searching large collections of graph data that has been implemented on a distributed platform and is in daily use for searching a database of molecules. The core of the approach uses an ‘inverted’ relaxation labelling method that finds a good match of the input data with stored examples. The method is shown to scale linearly with the number of graphs, and to scale linearly under given circumstances to the number of nodes in the graph. Key to the idea is that the system cuts the search time by removing a set of sub-optimal matches leaving those that could match. The system uses arrays of biologically plausible neural networks, Correlation Matrix Memories (CMMs) to store the constraints between the nodes of the graphs being searched. This is coupled to a novel search method. The system is highly parallel. Recently we have developed a parallel Grid enabled computer system (Cortex II) which utilises Digital Signal Processors (DSPs) and Field Programmable Gate Arrays (FPGAs) and have implemented the method on this system. A service for matching small molecules to a molecule database, in which the molecules are represented as attributed graphs, is currently running online. The methods have also been applied to searching trademark databases allowing people to find trademarks that are geometrically similar. The chapter describes the method in detail and its implementation and application. It also brings together work that has appeared separately and presents a new mathematical formulation of the mapping of RBE onto correlation matrix methods.


Field Programmable Gate Array Graph Match Database Graph Model Node Query 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.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Alwis, S., Austin, S.: A novel architecture for trade mark image retrieval systems. In: Mira, J. (ed.) IWANN 1999. LNCS, vol. 1607, pp. 361–372. Springer, Heidelberg (1999)CrossRefGoogle Scholar
  2. 2.
    Austin, J., Stonham, T.J.: Distributed associative memory for use in scene analysis. Image Vision Computing 5(4), 251–260 (1987)CrossRefGoogle Scholar
  3. 3.
    Austin, J., Kennedy, J., Lees, K.: A neural architecture for fast rule matching. In: Proceedings of the Artificial Neural Networks and Expert Systems Conference, Dunedin, New Zealand, pp. 255–260 (1995)Google Scholar
  4. 4.
    Austin, J.: RAM-based neural networks. World Scientific, River Edge (1998)Google Scholar
  5. 5.
    ACA group, AURA web pages (2008), (accessed January 1, 2009)
  6. 6.
    Austin, J., Davis, R., Fletcher, M., Jackson, T., Jessop, M., Liang, B., Pasley, A.: DAME: Searching large data sets within a grid-enabled engineering application. Proceedings of the IEEE - Special Issue on Grid Computing 93(3), 496–509 (2005)Google Scholar
  7. 7.
    Bermak, A., Austin, J.: VLSI implementation of a binary neural network - two case studies. In: Proceedings of the 7th International Conference on Microelectronics (Microneuro), p. 374. IEEE Computer Society Press, Los Alamitos (1999)Google Scholar
  8. 8.
    Bunke, H.: Graph matching: Theoretical foundations, algorithms, and applications. In: Proceedings of the International Conference on Vision Interface, pp. 82–88 (2000)Google Scholar
  9. 9.
    Bunke, H.: On a relation between graph edit distance and maximum common sub-graph. Pattern Recognition Letters 18(8), 689–694 (1997)CrossRefMathSciNetGoogle Scholar
  10. 10.
    Cybula, Molecular matcher web site (2008), (accessed January 1, 2009)
  11. 11.
    Fletcher, M., Jackson, T., Jessop, M., Liang, B., Austin, J.: The signal data explorer: A high performance grid based signal search tool for use in distributed diagnostic applications. In: Proceedings of the 6th IEEE International Symposium on Cluster Computing and the Grid, pp. 217–224. IEEE Computer Society, Los Alamitos (2006)Google Scholar
  12. 12.
    Hancock, E., Kittler, J.: Edge-labeling using dictionary-based relaxation. IEEE Transactions on Pattern Analysis and Machine Intelligence 12(2), 165–181 (1990)CrossRefGoogle Scholar
  13. 13.
    Hebb, D.O.: The organization of behavior. Wiley, New York (1949)Google Scholar
  14. 14.
    Klinger, S.: Chemical similarity searching with neural graph matching methods. PhD The University of York, UK (2006)Google Scholar
  15. 15.
    Lomas, D.: Improving automated postal address recognition using neural networks, PhD Thesis. The University of York, UK (2002)Google Scholar
  16. 16.
    Simpson, G.G.: Mammals and the nature of continents. American Journal of Science 241, 1–31 (1943)Google Scholar
  17. 17.
    Turner, M., Austin, J.: A neural relaxation technique for chemical graph matching. In: Niranjan, M. (ed.) Proceedings of the 5th International Conference on Artificial Neural Networks, pp. 7–9 (1997)Google Scholar
  18. 18.
    Tanimoto, T.T.: IBM Internal Report (November 1957)Google Scholar
  19. 19.
    Turner, M., Austin, J.: Matching performance of binary correlation matrix memories. Neural Networks 10(9), 1637–1648 (1997)CrossRefGoogle Scholar
  20. 20.
    Turner, M., Austin, J.: Graph matching by neural relaxation. Neural Computing and Applications 7(3), 238–248 (1998)CrossRefGoogle Scholar
  21. 21.
    Ullmann, J.R.: An algorithm for subgraph isomorphism. Journal of the ACM 23(1), 31–42 (1976)CrossRefMathSciNetGoogle Scholar
  22. 22.
    Waltz, D.: Understanding line drawings of scenes with shadows. In: Winston, P.H. (ed.) The Psychology of Computer Vision. McGraw-Hill, New York (1975)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2009

Authors and Affiliations

  • Jim Austin
    • 1
  1. 1.Advanced Computer Architectures Group, Department of Computer ScienceUniversity of YorkYorkUK

Personalised recommendations