Advertisement

Output-Error Model Training for Gaussian Process Models

  • Juš Kocijan
  • Dejan Petelin
Part of the Lecture Notes in Computer Science book series (LNCS, volume 6594)

Abstract

The training of a regression model depends on the purpose of the model. When a black-box model of dynamic systems is trained, two purposes are particularly common: prediction and simulation. The purpose of this paper is to highlight the differences between the learning of a dynamic-system model for prediction and for simulation in the presence of noise for Gaussian process models. Gaussian process models are probabilistic, nonparametric models that recently generated interest in the machine-learning community. This method can also be used also for the modelling of dynamic systems, which is the main interest of the engineering community. The paper elaborates the differences between prediction- and simulation-purposed modelling in the presence of noise, which is more difficult in the case when we train the model for simulation. An example is given to illustrate the described differences.

Keywords

Gaussian process models dynamic systems regression autoregressive models output-error models 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Ažman, K., Kocijan, J.: Application of Gaussian processes for black-box modelling of biosystems. ISA Transactions 46, 443–457 (2007)CrossRefGoogle Scholar
  2. 2.
    Gibbs, M.N.: Bayesian Gaussian Processes for Regression and Classification. Ph.d. thesis, Cambridge University (1997)Google Scholar
  3. 3.
    Henson, M.A., Seborg, D.E.: Adaptive nonlinear control of a pH neutralization process. IEEE Transactions on Control System Technology 2, 169–183 (1994)CrossRefGoogle Scholar
  4. 4.
    Kocijan, J., Girard, A., Banko, B., Murray-Smith, R.: Dynamic systems identification with Gaussian processes. Mathematical and Computer Modelling of Dynamic Systems 11(4), 411–424 (2005)MathSciNetCrossRefzbMATHGoogle Scholar
  5. 5.
    Kocijan, J., Likar, B.: Gas-liquid separator modelling and simulation with Gaussian-process models. Simulation Modelling Practice and Theory 16(8), 910–922 (2008)CrossRefGoogle Scholar
  6. 6.
    Ljung, L.: System identification – theory for the user, 2nd edn. Prentice Hall, New Jersey (1999)zbMATHGoogle Scholar
  7. 7.
    Murray-Smith, R., Girard, A.: Gaussian Process priors with ARMA noise models. In: Proceedings of Irish Signals and Systems Conference, Maynooth, pp. 147–152 (2001)Google Scholar
  8. 8.
    Nelles, O.: Nonlinear System Identification. Springer, Heidelberg (2001)CrossRefzbMATHGoogle Scholar
  9. 9.
    Rasmusen, C.E., Williams, C.K.I.: Gaussian Processes for Machine Learning. MIT Press, Cambridge (2006)Google Scholar
  10. 10.
    Sjoeberg, J., Zhang, Q., Ljung, L., Benveniste, A., Delyon, B., Glorennec, P.-Y., Hjalmarsson, H., Juditsky, A.: Nonlinear black-box modelling in system identification: A unified overview. Automatica 31(12), 1691–1724 (1995)MathSciNetCrossRefzbMATHGoogle Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2011

Authors and Affiliations

  • Juš Kocijan
    • 1
    • 2
  • Dejan Petelin
    • 1
  1. 1.Jozef Stefan InstituteLjubljanaSlovenia
  2. 2.University of Nova GoricaNova GoricaSlovenia

Personalised recommendations