Advertisement

Analyzing bandit-based adaptive operator selection mechanisms

  • Álvaro Fialho
  • Luis Da Costa
  • Marc SchoenauerEmail author
  • Michèle Sebag
Article

Abstract

Several techniques have been proposed to tackle the Adaptive Operator Selection (AOS) issue in Evolutionary Algorithms. Some recent proposals are based on the Multi-armed Bandit (MAB) paradigm: each operator is viewed as one arm of a MAB problem, and the rewards are mainly based on the fitness improvement brought by the corresponding operator to the individual it is applied to. However, the AOS problem is dynamic, whereas standard MAB algorithms are known to optimally solve the exploitation versus exploration trade-off in static settings. An original dynamic variant of the standard MAB Upper Confidence Bound algorithm is proposed here, using a sliding time window to compute both its exploitation and exploration terms. In order to perform sound comparisons between AOS algorithms, artificial scenarios have been proposed in the literature. They are extended here toward smoother transitions between different reward settings. The resulting original testbed also includes a real evolutionary algorithm that is applied to the well-known Royal Road problem. It is used here to perform a thorough analysis of the behavior of AOS algorithms, to assess their sensitivity with respect to their own hyper-parameters, and to propose a sound comparison of their performances.

Keywords

Parameter control Adaptive Operator Selection Multi-Armed Bandits 

Mathematics Subject Classifications (2010)

68T05 68W40 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Auer, P., Cesa-Bianchi, N., Fischer, P.: Finite-time analysis of the multi-armed bandit problem. Mach. Learn. 47(2–3), 235–256 (2002)zbMATHCrossRefGoogle Scholar
  2. 2.
    Barbosa, H.J.C., Sá, A.M.: On adaptive operator probabilities in real coded genetic algorithms. In: XX Intl. Conference of the Chilean Computer Science Society (2000)Google Scholar
  3. 3.
    Bartz-Beielstein, T., Lasarczyk, C., Preuss, M.: Sequential parameter optimization. In: McKay, B. (ed.) Proc. Congress on Evolutionary Computation, pp. 773–780. IEEE (2005)Google Scholar
  4. 4.
    Birattari, M., Stützle, T., Paquete, L., Varrentrapp, K.: A racing algorithm for configuring metaheuristics. In: Langdon, W.B., et al. (eds.) Proc. Genetic and Evolutionary Computation Conference, pp. 11–18. Morgan Kaufmann (2002)Google Scholar
  5. 5.
    Collet, P., Schoenauer, M.: GUIDE: unifying evolutionary engines through a graphical user interface. In: Liardet, P., et al. (eds.) Proc. Intl. Conference on Artificial Evolution. LNCS, vol. 2936, pp. 203–215. Springer (2003)Google Scholar
  6. 6.
    Conover, W.J.: Practical Nonparametric Statistics. Wiley (1999)Google Scholar
  7. 7.
    Da Costa, L., Fialho, A., Schoenauer, M., Sebag, M.: Adaptive operator selection with dynamic multi-armed bandits. In: Keijzer, M., et al. (eds.) Proc. Genetic and Evolutionary Computation Conference, pp. 913–920. ACM (2008)Google Scholar
  8. 8.
    Davis, L.: Adapting operator probabilities in genetic algorithms. In: Schaffer, J.D. (ed.) Proc. Intl. Conference on Genetic Algorithms, pp. 61–69. Morgan Kaufmann (1989)Google Scholar
  9. 9.
    DeJong, K.: Evolutionary Computation. A unified Approach. MIT (2006)Google Scholar
  10. 10.
    DeJong, K.: Parameter setting in EAs: a 30 year perspective. In: Lobo, F., Lima, C., Michalewicz, Z. (eds.): Parameter Setting in Evolutionary Algorithms. Studies in Computational Intelligence, vol. 54, pp. 1–18. Springer (2007)Google Scholar
  11. 11.
    Eiben, A.E., Hinterding, R., Michalewicz, Z.: Parameter control in Evolutionary Algorithms. IEEE Trans. Evol. Comput. 3(2), 124–141 (1999)CrossRefGoogle Scholar
  12. 12.
    Eiben, A.E., Michalewicz, Z., Schoenauer, M., Smith, J.E.: Parameter control in evolutionary algorithms. In: Lobo, F., Lima, C., Michalewicz, Z. (eds.): Parameter Setting in Evolutionary Algorithms. Studies in Computational Intelligence, vol. 54, pp. 19–46. Springer (2007)Google Scholar
  13. 13.
    Eiben, A.E., Smith, J.E.: Introduction to Evolutionary Computing. Springer (2003)Google Scholar
  14. 14.
    Fialho, A., Da Costa, L., Schoenauer, M., Sebag, M.: Extreme value based adaptive operator selection. In: Rudolph, G., et al. (eds.) Proc. Intl. Conference on Parallel Solving from Nature. LNCS, vol. 5199, pp. 175–184. Springer (2008)Google Scholar
  15. 15.
    Fialho, A., Da Costa, L., Schoenauer, M., Sebag, M.: Dynamic multi-armed bandits and extreme value-based rewards for adaptive operator selection in evolutionary algorithms. In: Stützle, T. (ed.) Proc. 3rd Intl. Conference on Learning and Intelligent Optimization. LNCS, vol. 5851, pp. 176–190. Springer (2009)Google Scholar
  16. 16.
    Fialho, A., Schoenauer, M., Sebag, M.: Analysis of adaptive operator selection techniques on the royal road and long k-path problems. In: Raidl, G., et al. (eds.) Proc. Genetic and Evolutionary Computation Conference, pp. 779–786. ACM (2009)Google Scholar
  17. 17.
    Fogel, D.B.: Phenotypes, genotypes and operators in evolutionary computation. In: Proc. Intl. Conference on Evolutionary Computation. IEEE (1995)Google Scholar
  18. 18.
    Gagliolo, M., Schmidhuber, J.: Algorithm Selection as a Bandit Problem with Unbounded Losses. Tech. Rep. IDSIA-07-08, IDSIA (2008)Google Scholar
  19. 19.
    Goldberg, D.: Probability matching, the magnitude of reinforcement, and classifier system bidding. Mach. Learn. 5(4), 407–426 (1990)Google Scholar
  20. 20.
    Gould, S., Eldredge, N.: Punctuated equilibria: the tempo and mode of evolution reconsidered. Paleobiology 3(2), 115–151 (1977)Google Scholar
  21. 21.
    Hartland, C., Baskiotis, N., Gelly, S., Teytaud, O., Sebag, M.: Change point detection and meta-bandits for online learning in dynamic environments. In: Proc. Conférence Francophone sur l’Apprentissage Automatique (2007)Google Scholar
  22. 22.
    Hartland, C., Gelly, S., Baskiotis, N., Teytaud, O., Sebag, M.: Multi-armed bandit, dynamic environments and meta-bandits. In: Online Trading of Exploration and Exploitation Workshop, NIPS (2006)Google Scholar
  23. 23.
    Hinkley, D.: Inference about the change point from cumulative sum-tests. Biometrika 58(3), 509–523 (1970)CrossRefMathSciNetGoogle Scholar
  24. 24.
    Holland, J.H.: Royal road functions. In: Internet Genetic Algorithms Digest, vol. 7, p. 22. Massachusetts Institute of Technology (1993)Google Scholar
  25. 25.
    Jones, T.: A description of Holland’s Royal Road. Evol. Comput. 2(4), 409–415 (1994)CrossRefGoogle Scholar
  26. 26.
    Julstrom, B.: What have you done for me lately? Adapting operator probabilities in a steady-state genetic algorithm. In: Eshelman, L.J., et al. (eds.) Proc. Intl. Conference on Genetic Algorithms, pp. 81–87. Morgan Kaufmann (1995)Google Scholar
  27. 27.
    Kallel, L., Schoenauer, M.: Fitness Distance Correlation for Variable Length Representations. Tech. Rep. 363, CMAP, Ecole Polytechnique (1996)Google Scholar
  28. 28.
    Lai, T., Robbins, H.: Asymptotically efficient adaptive allocation rules. Adv. Appl. Math. 6(1), 4–22 (1985)zbMATHCrossRefMathSciNetGoogle Scholar
  29. 29.
    Lobo, F., Goldberg, D.: Decision making in a hybrid genetic algorithm. In: Porto, B. (ed.) Proc. Intl. Conference on Evolutionary Computation, pp. 121–125. IEEE (1997)Google Scholar
  30. 30.
    Lobo, F., Lima, C., Michalewicz, Z. (eds.): Parameter Setting in Evolutionary Algorithms. Studies in Computational Intelligence, vol. 54. Springer (2007)Google Scholar
  31. 31.
    Maturana, J., Fialho, A., Saubion, F., Schoenauer, M., Sebag, M.: Extreme compass and dynamic multi-armed bandits for adaptive operator selection. In: Proc. Congress on Evolutionary Computation, pp. 365–372. IEEE (2009)Google Scholar
  32. 32.
    Maturana, J., Lardeux, F., Saubion, F.: Autonomous operator management for evolutionary algorithms. Journal of Heuristics (2010). doi:  10.1007/s10732-010-9125-3 Google Scholar
  33. 33.
    Maturana, J., Saubion, F.: A compass to guide genetic algorithms. In: Rudolph, G., et al. (eds.) Proc. Intl. Conference on Parallel Solving from Nature. LNCS, vol. 5199, pp. 256–265. Springer (2008)Google Scholar
  34. 34.
    Michalewicz, Z.: Genetic Algorithms + Data Structures = Evolution Programs, 3rd edn. Springer, New York (1996)zbMATHGoogle Scholar
  35. 35.
    Mitchell, M., Forrest, S., Holland, J.H.: The royal road for genetic algorithms: fitness landscapes and GA performance. In: Proc. European Conference on Artificial Life, pp. 245–254 (1992)Google Scholar
  36. 36.
    Nannen, V., Eiben, A.E.: Relevance estimation and value calibration of evolutionary algorithm parameters. In: Veloso, M. (ed.) Proc. Intl. Joint Conference on Artificial Intelligence, pp. 975–980 (2007)Google Scholar
  37. 37.
    Quick, R.J., Rayward-Smith, V.J., Smith, G.D.: The royal road functions: description, intent and experimentation. In: Selected Papers from AISB Workshop on Evolutionary Computing. LNCS, vol. 1143, pp. 223–235. Springer (1996)Google Scholar
  38. 38.
    Spears, W.: Adapting crossover in evolutionary algorithms. In: McDonnell, J.R., et al. (eds.) Proc. Conference on Evolutionary Programming, pp. 367–384. MIT (1995)Google Scholar
  39. 39.
    Stützle, T. (ed.): Proc. 3rd Intl. Conference on Learning and Intelligent Optimization. LNCS, vol. 5851. Springer (2009)Google Scholar
  40. 40.
    Thierens, D.: An adaptive pursuit strategy for allocating operator probabilities. In: Beyer, H.G. (eds.) Proc. Genetic and Evolutionary Computation Conference, pp. 1539–1546. ACM (2005)Google Scholar
  41. 41.
    Thierens, D.: Adaptive strategies for operator allocation. In: Lobo, F., Lima, C., Michalewicz, Z. (eds.): Parameter Setting in Evolutionary Algorithms. Studies in Computational Intelligence, vol. 54, pp. 77–90. Springer (2007)Google Scholar
  42. 42.
    Tuson, A., Ross, P.: Adapting operator settings in genetic algorithms. Evol. Comput. 6(2), 161–184 (1998)CrossRefGoogle Scholar
  43. 43.
    Whitacre, J., Pham, T., Sarker, R.: Use of statistical outlier detection method in adaptive evolutionary algorithms. In: Keijzer, M. (ed.) Proc. Genetic and Evolutionary Computation Conference, pp. 1345–1352. ACM (2006)Google Scholar
  44. 44.
    Yu, T., Davis, D., Baydar, C., Roy, R. (eds.): Evolutionary Computation in Practice. Studies in Computational Intelligence, vol. 88. Springer (2008)Google Scholar
  45. 45.
    Yuan, B., Gallagher, M.: Statistical racing techniques for improved empirical evaluation of evolutionary algorithms. In: Yao, X., et al. (eds.) Proc. Intl. Conference on Parallel Solving from Nature. LNCS, vol. 3242, pp. 172–181. Springer (2004)Google Scholar

Copyright information

© Springer Science+Business Media B.V. 2010

Authors and Affiliations

  • Álvaro Fialho
    • 1
  • Luis Da Costa
    • 2
  • Marc Schoenauer
    • 1
    • 2
    Email author
  • Michèle Sebag
    • 1
    • 2
  1. 1.Microsoft Research–INRIA Joint CentreOrsay CedexFrance
  2. 2.Project-Team TAOINRIA Saclay—Île-de-France & LRI (UMR CNRS 8623)Orsay CedexFrance

Personalised recommendations