Abstract
A simple model shows how a reasonable update scheme for the probability vector by which a hyper-heuristic chooses the next heuristic leads to neglecting useful mutation heuristics. Empirical evidence supports this on the MaxSat, TravelingSalesman, PermutationFlowshop and VehicleRoutingProblem problems. A new approach to hyper-heuristics is proposed that addresses this problem by modeling and learning hyper-heuristics by means of a hidden Markov Model. Experiments show that this is a feasible and promising approach.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
See goo.gl/vVTZNE for details.
- 2.
If the heuristics are applied with uniform probability, around \(5\,\%\) to \(20\,\%\) of the time, although it strongly depends on the problem. See goo.gl/vVTZNE for empirical evidence.
- 3.
See the right axis for the appropriate unit.
- 4.
Experiments show that such local optimum is nearly always near the global optimum, although sequences can be derived that are hard to learn.
- 5.
This in contrast with the Markov assumption that states that a Markov process has no memory.
- 6.
See Sect. 5.3.
- 7.
The decision component will however have an impact on the generated evidence.
- 8.
Since the Baum-Welch algorithm is a heuristic learning algorithm, this is not guaranteed, but we never encountered such an example.
- 9.
For instance, how a hyper-heuristic can model a local search heuristic.
References
Khamassi, I.: Ant-Q hyper heuristic approach applied to the cross-domain heuristic search challenge problems. In: CHeSC 2011 (2011)
McClymont, K., Keedwell, E.: A single objective variant of the online selective Markov chain Hyper-heuristic (MCHH-S). In: CHeSC 2011 (2011)
Wikipedia: Probability vector – wikipedia, the free encyclopedia (2014). Accessed 24 Sep 2014
Lipowski, A., Lipowska, D.: Roulette-wheel selection via stochastic acceptance. Phys. A Stat. Mech. Appl. 391(6), 2193–2196 (2012)
Ochoa, G., et al.: HyFlex: a benchmark framework for cross-domain heuristic search. In: Hao, J.-K., Middendorf, M. (eds.) EvoCOP 2012. LNCS, vol. 7245, pp. 136–147. Springer, Heidelberg (2012)
Watkins, C.J.C.H.: Learning from delayed rewards. Ph.D. thesis, King’s College, Cambridge, UK, May 1989
Gaspero, L.D., Urli, T.: A reinforcement learning approach for the cross-domain heuristic search challenge. In: CHeSC 2011 (2011)
Mısır, M., De Causmaecker, P., Vanden Berghe, G., Verbeeck, K.: An adaptive hyper-heuristic for CHeSC 2011. In: CHeSC 2011 (2011)
Mısır, M., Verbeeck, K., De Causmaecker, P., Vanden Berghe, G.: An intelligent hyper-heuristic framework for CHeSC 2011. In: Hamadi, Y., Schoenauer, M. (eds.) LION 2012. LNCS, vol. 7219, pp. 461–466. Springer, Heidelberg (2012)
Glover, F., Hanafi, S.: Tabu search and finite convergence. Discrete Appl. Math. 119(1–2), 3–36 (2002)
Burke, E.K., Curtois, T., Hyde, M.R., Kendall, G., Ochoa, G., Petrovic, S., RodrÃguez, J.A.V., Gendreau, M.: Iterated local search vs. hyper-heuristics: towards general-purpose search algorithms. In: IEEE Congress on Evolutionary Computation, pp. 1–8. IEEE (2010)
Bause, F.: Input-output hidden markov models for the aggregation of performance models. Sfb Teilprojekt M and Ls Informatik Iv and Modellierung Grosser and Sfb Teilprojekt M SFB559-03010, Universität Dortmund, Dortmund, July 2003
Markovitch, S., Scott, P.D.: The role of forgetting in learning. In: Laird, J.E. (ed.) ML, pp. 459–465. Morgan Kaufmann, Ann Arbor (1988)
Asta, S., Özcan, E., Parkes, A.J.: Batched mode hyper-heuristics. In: Nicosia, G., Pardalos, P. (eds.) LION 7. LNCS, vol. 7997, pp. 404–409. Springer, Heidelberg (2013)
Van Onsem, W., Demoen, B., De Causmaecker, P.: HHaaHHM: hyper-heuristics as a hidden Markov model. In: Proceedings of the Cross-domain Heuristic Selection Competition 2014, April 2014
Pihera, J., Musliu, N.: ETHH - evolving tree hyper-heuristic. In: Proceedings of the Cross-domain Heuristic Selection Competition 2014, April 2014
Acknowledgements
This research is funded by the Institute for Innovation through Science and Technology (IWT) under grant \(131'751\).
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2015 Springer International Publishing Switzerland
About this paper
Cite this paper
Van Onsem, W., Demoen, B., De Causmaecker, P. (2015). Learning a Hidden Markov Model-Based Hyper-heuristic. In: Dhaenens, C., Jourdan, L., Marmion, ME. (eds) Learning and Intelligent Optimization. LION 2015. Lecture Notes in Computer Science(), vol 8994. Springer, Cham. https://doi.org/10.1007/978-3-319-19084-6_7
Download citation
DOI: https://doi.org/10.1007/978-3-319-19084-6_7
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-319-19083-9
Online ISBN: 978-3-319-19084-6
eBook Packages: Computer ScienceComputer Science (R0)