Learning Inherent Networks from Stochastic Search Methods

  • David Iclănzan
  • Fabio Daolio
  • Marco Tomassini
Part of the Lecture Notes in Computer Science book series (LNCS, volume 8600)

Abstract

Analysis and modeling of search heuristics operating on complex problems is a difficult albeit important research area. Inherent networks, i.e. the graphs whose vertices represent local optima and the edges describe the weighted transition probabilities between them, enable a network characterization of combinatorial fitness landscapes. Methods revealing such inherent structures of the search spaces in relation to deterministic move operators, have been recently developed for small problem instances. This work proposes a more general, scalable, data-driven approach, that extracts the transition probabilities from actual runs of metaheuristics, capturing the effect and interplay of a broader spectrum of factors. Using the case of NK landscapes, we show that such an unsupervised learning approach is successful in quickly providing a coherent view of the inherent network of a problem instance.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Hoos, H.H., Stützle, T.: Stochastic Local Search: Foundations and Applications. Morgan Kaufmann, San Francisco (2005)Google Scholar
  2. 2.
    Watson, J.P.: An introduction to fitness landscape analysis and cost models for local search. In: Handbook of Metaheuristics, pp. 599–623. Springer (2010)Google Scholar
  3. 3.
    Tomassini, M., Verel, S., Ochoa, G.: Complex-network analysis of combinatorial spaces: The NK landscape case. Physical Review E 78(6), 066114 (2008)Google Scholar
  4. 4.
    Verel, S., Ochoa, G., Tomassini, M.: Local optima networks of NK landscapes with neutrality. IEEE Transactions on Evolutionary Computation 15(6), 783–797 (2011)CrossRefGoogle Scholar
  5. 5.
    Doye, J.P.K.: The network topology of a potential energy landscape: a static scale-free network. Phys. Rev. Lett. 88, 238701 (2002)Google Scholar
  6. 6.
    Vérel, S., Daolio, F., Ochoa, G., Tomassini, M.: Local optima networks with escape edges. In: Hao, J.-K., Legrand, P., Collet, P., Monmarché, N., Lutton, E., Schoenauer, M. (eds.) EA 2011. LNCS, vol. 7401, pp. 49–60. Springer, Heidelberg (2012)CrossRefGoogle Scholar
  7. 7.
    Daolio, F., Verel, S., Ochoa, G., Tomassini, M.: Local optima networks and the performance of iterated local search. In: GECCO, pp. 369–376. ACM (2012)Google Scholar
  8. 8.
    Ochoa, G., Verel, S., Tomassini, M.: First-improvement vs. Best-improvement local optima networks of NK landscapes. In: Schaefer, R., Cotta, C., Kołodziej, J., Rudolph, G. (eds.) PPSN XI. LNCS, vol. 6238, pp. 104–113. Springer, Heidelberg (2010)CrossRefGoogle Scholar
  9. 9.
    Kauffman, S.: The origins of order: Self organization and selection in evolution. Oxford University Press (1993)Google Scholar
  10. 10.
    Kohonen, T., Schroeder, M.R., Huang, T.S. (eds.): Self-Organizing Maps, 3rd edn. Springer-Verlag New York, Inc., Secaucus (2001)MATHGoogle Scholar
  11. 11.
    Romero, G., Arenas, M.G., Castillo, P., Merelo, J.: Visualization of neural net evolution. In: Mira, J., Álvarez, J.R. (eds.) IWANN 2003. LNCS, vol. 2686, pp. 534–541. Springer, Heidelberg (2003)CrossRefGoogle Scholar
  12. 12.
    Smith-Miles, K., Lopes, L.: Generalising algorithm performance in instance space: A timetabling case study. In: Coello, C.A.C. (ed.) LION 5. LNCS, vol. 6683, pp. 524–538. Springer, Heidelberg (2011)CrossRefGoogle Scholar
  13. 13.
    Amor, H.B., Rettinger, A.: Intelligent exploration for genetic algorithms: using self-organizing maps in evolutionary computation. In: GECCO, pp. 1531–1538. ACM (2005)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2014

Authors and Affiliations

  • David Iclănzan
    • 1
  • Fabio Daolio
    • 1
  • Marco Tomassini
    • 1
  1. 1.Faculty of Business and EconomicsUniversity of LausanneSwitzerland

Personalised recommendations