Theory and Applications of Hybrid Simulated Annealing

Part of the Intelligent Systems Reference Library book series (ISRL, volume 38)


Local optimization techniques such as gradient-based methods and the expectation-maximization algorithm have an advantage of fast convergence but do not guarantee convergence to the global optimum. On the other hand, global optimization techniques based on stochastic approaches such as evolutionary algorithms and simulated annealing provide the possibility of global convergence, which is accomplished at the expense of computational and time complexity. This chapter aims at demonstrating how these two approaches can be effectively combined for improved convergence speed and quality of the solution. In particular, a hybrid method, called hybrid simulated annealing (HSA), is presented, where a simulated annealing algorithm is combined with local optimization methods. First, its general procedure and mathematical convergence properties are described. Then, its two example applications are presented, namely, optimization of hidden Markov models for visual speech recognition and optimization of radial basis function networks for pattern classification, in order to show how the HSA algorithm can be successfully adopted for solving real-world problems effectively. As an appendix, the source code for multi-dimensional Cauchy random number generation is provided, which is essential for implementation of the presented method.


Hide Markov Model Speech Recognition Radial Basis Function Network Annealing Schedule Local Optimization Method 
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.
    Bahl, L.R., Brown, P.F., de Souza, P.V., Mercer, R.L.: Maximum mutual information estimation of hidden Markov model parameters for speech recognition. In: Proc. Int. Conf. Acoustics, Speech and Signal Processing, Tokyo, Japan, pp. 49–52 (1986)Google Scholar
  2. 2.
    Ben-Yishai, A., Burshtein, D.: A discriminative training algorithm for hidden Markov models. IEEE Trans. Speech Audio Processing 12(3), 204–216 (2004)CrossRefGoogle Scholar
  3. 3.
    Benoudjit, N., Archambeau, C., Lendasse, A., Lee, M.V.J.A.: Width optimization of the Gaussian kernels in radial basis function networks. In: Proc. European Symposium on Artificial Neural Networks, Bruges, Belgium, pp. 425–432 (2002)Google Scholar
  4. 4.
    Bishop, C.M.: Neural Networks for Pattern Recognition. Oxford Univ. Press, Inc., New York (1995)Google Scholar
  5. 5.
    Blake, C.L., Merz, C.J.: UCI repository of machine learning database. Dept. Information and Computer Science. Univ. California (1998)Google Scholar
  6. 6.
    Chen, S., Cowan, C.F.N., Grant, P.M.: Orthogonal least squares learning algorithm for radial basis function networks. IEEE Trans. Neural Networks 2(2), 302–309 (1991)CrossRefGoogle Scholar
  7. 7.
    Chen, T., Chen, H.: Approximation capability to functions of several variables, nonlinear functionals, and operators by radial basis function networks. IEEE Trans. Neural Networks 6, 904–910 (1995)CrossRefGoogle Scholar
  8. 8.
    Chibelushi, C.C., Deravi, F., Mason, J.S.D.: A review of speech-based bimodal recognition. IEEE Trans. Multimedia 4(1), 23–37 (2002)CrossRefGoogle Scholar
  9. 9.
    Dempster, A.P., Laird, N.M., Rubin, D.B.: Maximum likelihood from incomplete data via the EM algorithm. J. R. Statist. Soc., Ser. B 39(1), 1–38 (1977)MathSciNetzbMATHGoogle Scholar
  10. 10.
    Geman, S., Geman, D.: Stochastic relaxation, Gibbs distributions, and the Bayesian restoration of images. IEEE Trans. Pattern Anal. Machine Intell. 6(6), 721–741 (1984)zbMATHCrossRefGoogle Scholar
  11. 11.
    Ingber, L., Rosen, B.: Genetic algorithms and very fast simulated reannealing: a comparison. Mathematical and Computer Modeling 16(11), 87–100 (1992)MathSciNetzbMATHCrossRefGoogle Scholar
  12. 12.
    Juang, B.H., Chou, W., Lee, C.H.: Minimum classification error rate methods for speech recognition. IEEE Trans. Speech Audio Processing 5(3), 257–265 (1997)CrossRefGoogle Scholar
  13. 13.
    Kirkpatrick, S., Gerlatt, C.D., Vecchi, M.P.: Optimization by simulated annealing. Science 220, 671–680 (1983)MathSciNetzbMATHCrossRefGoogle Scholar
  14. 14.
    Laarhoven, P.J.M.V., Aarts, E.H.L.: Simulated Annealing: Theory and Applications. Kluwer Academic Publishers, Dordrecht (1987)zbMATHGoogle Scholar
  15. 15.
    Lee, J.S., Park, C.H.: Robust audio-visual speech recognition based on late integration. IEEE Trans. Multimedia 10(5), 767–779 (2008)CrossRefGoogle Scholar
  16. 16.
    Lee, J.S., Park, C.H.: Global optimization of radial basis function networks by hybrid simulated annealing. Neural Network World 20(4), 519–537 (2010a)zbMATHGoogle Scholar
  17. 17.
    Lee, J.S., Park, C.H.: Hybrid simulated annealig and its application to optimization of hidden Markov models for visual speech recognition. IEEE Trans. Syst., Man, Cybern. B 40(4), 1188–1196 (2010b)CrossRefGoogle Scholar
  18. 18.
    Marsaglia, G.: Choosing a point from the surface of a sphere. Ann. Math. Statist. 43(2), 645–646 (1972)zbMATHCrossRefGoogle Scholar
  19. 19.
    Metropolis, N., Rosenbluth, A.W., Rosenbluth, M.N., Teller, A.H., Teller, E.: Equation of state calculations by fast computing machines. J. Chem. Phys. 21(6), 1087–1092 (1953)CrossRefGoogle Scholar
  20. 20.
    Mitra, D., Romeo, F., Sangiovanni-Vincentelli, A.: Convergence and finite-time behavior of simulated annealing. Advances in Applied Probability 18, 747–771 (1986)MathSciNetzbMATHCrossRefGoogle Scholar
  21. 21.
    Moody, J., Darken, C.: Fast learning in networks of locally-tuned processing units. Neural Computation 1, 291–294 (1989)CrossRefGoogle Scholar
  22. 22.
    Nam, D., Lee, J.S., Park, C.H.: n-dimensional Cauchy neighbor generation for the fast simulated annealing. IEICE Trans. Inf. Syst. E87-D(11), 2499–2502 (2004)Google Scholar
  23. 23.
    Orr, M.J.L.: Regularisation in the selection of radial basis function centers. Neural Computation 7, 606–623 (1995)CrossRefGoogle Scholar
  24. 24.
    Orr, M.J.L.: Introduction to radial basis function networks. Tech. rep., Center for Cognitive Science, Univ. Edinburgh (1996)Google Scholar
  25. 25.
    Park, J., Sandberg, I.W.: Approximation and radial basis function networks. Neural Computation 5, 305–316 (1993)CrossRefGoogle Scholar
  26. 26.
    Paul, D.: Training of HMM recognizers by simulated annealing. In: Proc. Int. Conf. Acoustics, Speech and Signal Processing, Tampa, FL, pp. 13–16 (1985)Google Scholar
  27. 27.
    Rabiner, L., Juang, B.H.: Fundamentals of Speech Recognition. Prentice Hall, Englewood Cliffs (1993)Google Scholar
  28. 28.
    Rao, A.V., Rose, K.: Deterministically annealed design of hidden Markov model speech recognizers. IEEE Trans. Speech Audio Processing 9(2), 111–126 (2001)CrossRefGoogle Scholar
  29. 29.
    Rodrígues, L.J., Torres, I.: Comparative Study of the Baum-Welch and Viterbi Training Algorithms Applied to Read and Spontaneous Speech Recognition. In: Perales, F.J., Campilho, A.C., Pérez, N., Sanfeliu, A. (eds.) IbPRIA 2003. LNCS, vol. 2652, pp. 847–857. Springer, Heidelberg (2003)CrossRefGoogle Scholar
  30. 30.
    Schwenker, F., Kestler, H.A., Palm, G.: Three learning phases for radial-basis-function networks. Neural Networks 14, 439–458 (2001)CrossRefGoogle Scholar
  31. 31.
    Szu, H.H., Hartley, R.L.: Fast simulated annealing. Phys. Lett. A 122(3-4), 157–162 (1987)CrossRefGoogle Scholar
  32. 32.
    Yang, R.L.: Convergence of the simulated annealing algorithm for continuous global optimization. J. Optimization Theory and Applications 104(3), 691–716 (2000)zbMATHCrossRefGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2013

Authors and Affiliations

  1. 1.School of Integrated TechnologyYonsei UniversityIncheonKorea
  2. 2.Department of Electrical EngineeringKAISTDaejeonKorea
  3. 3.Institute of Electrical EngineeringSwiss Federal Institute of Technology Lausanne (EPFL)LausanneSwitzerland

Personalised recommendations