Advertisement

Efficient Domain Decomposition for a Neural Network Learning Algorithm, Used for the Dose Evaluation in External Radiotherapy

  • Marc Sauget
  • Rémy Laurent
  • Julien Henriet
  • Michel Salomon
  • Régine Gschwind
  • Sylvain Contassot-Vivier
  • Libor Makovicka
  • Charles Soussen
Part of the Lecture Notes in Computer Science book series (LNCS, volume 6352)

Abstract

The purpose of this work is to further study the relevance of accelerating the Monte Carlo calculations for the gamma rays external radiotherapy through feed-forward neural networks. We have previously presented a parallel incremental algorithm that builds neural networks of reduced size, while providing high quality approximations of the dose deposit. Our parallel algorithm consists in a regular decomposition of the initial learning dataset (also called learning domain) in as much subsets as available processors. However, the initial learning set presents heterogeneous signal complexities and consequently, the learning times of regular subsets are very different. This paper presents an efficient learning domain decomposition which balances the signal complexities across the processors. As will be shown, the resulting irregular decomposition allows for important gains in learning time of the global network.

Keywords

Domain Decomposition Beam Width External Radiotherapy Monte Carlo Calculation Learning Time 
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.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Bahi, J., Contassot-Vivier, S., Sauget, M., Vasseur, A.: A parallel incremental learning algorithm for neural networks with fault tolerance. In: Palma, J.M.L.M., Amestoy, P.R., Daydé, M., Mattoso, M., Lopes, J.C. (eds.) VECPAR 2008. LNCS, vol. 5336, pp. 502–515. Springer, Heidelberg (2008)Google Scholar
  2. 2.
    Bahi, J.M., Contassot-Vivier, S., Makovicka, L., Martin, E., Sauget, M.: Neurad. Agence pour la Protection des Programmes. No: IDDN.FR.001.130035.000.S.P.2006.000.10000 (2006)Google Scholar
  3. 3.
    Bahi, J.M., Contassot-Vivier, S., Makovicka, L., Martin, E., Sauget, M.: Neural network based algorithm for radiation dose evaluation in heterogeneous environments. In: Kollias, S.D., Stafylopatis, A., Duch, W., Oja, E. (eds.) ICANN 2006. LNCS, vol. 4132, pp. 777–787. Springer, Heidelberg (2006)CrossRefGoogle Scholar
  4. 4.
    Bahi, J.M., Contassot-Vivier, S., Sauget, M.: An incremental learning algorithm for functional approximation. Advances in Engineering Software 40(8), 725–730 (2009) doi:10.1016/j.advengsoft.2008.12.018MATHCrossRefGoogle Scholar
  5. 5.
  6. 6.
    Buffard, E., Gschwind, R., Makovicka, L., David, C.: Monte Carlo calculations of the impact of a hip prosthesis on the dose distribution. Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms 251(1), 9–18 (2006)CrossRefGoogle Scholar
  7. 7.
    Jones, M.T., Plassmann, P.E.: Computational results for parallel unstructured mesh computations (1994)Google Scholar
  8. 8.
    Jones, M.T., Plassmann, P.E.: Parallel algorithms for the adaptive refinement and partitioning of unstructured meshes. In: Proceedings of the Scalable High-Performance Computing Conference, pp. 478–485. IEEE, Los Alamitos (1997)Google Scholar
  9. 9.
    Makovicka, L., Vasseur, A., Sauget, M., Martin, E., Gschwind, R., Henriet, J., Salomon, M.: Avenir des nouveaux concepts des calculs dosimétriques basés sur les méthodes de Monte Carlo. Radioprotection 44(1), 77–88 (2009)CrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2010

Authors and Affiliations

  • Marc Sauget
    • 1
  • Rémy Laurent
    • 1
  • Julien Henriet
    • 1
  • Michel Salomon
    • 2
  • Régine Gschwind
    • 1
  • Sylvain Contassot-Vivier
    • 3
  • Libor Makovicka
    • 1
  • Charles Soussen
    • 4
  1. 1.Femto-ST, ENISYS/IRMAMontbéliardFrance
  2. 2.LIFC/ANDUniversity of Franche-ComtéBelfortFrance
  3. 3.LORIAUniversity of NancyVandoeuvre-lès-Nancy CedexFrance
  4. 4.Faculté des sciences et techniquesCRANVandoeuvre-lès-Nancy CedexFrance

Personalised recommendations