A New Training Algorithm for Neuro-Fuzzy Networks

Stefan Jakubek, Nikolaus Keuth



In this paper a new iterative construction algorithm for local model networks is presented. The algorithm is focussed on building models with sparsely distributed data as they occur in engine optimization processes. The validity function of each local model is fitted to the available data using statistical criteria along with regularisation and thus allowing an arbitrary orientation and extent in the input space. Local models are consecutively placed into those regions of the input space where the model error is still large thus guaranteeing maximal improvement through each new local model. The orientation and extent of each validity function is also adapted to the available training data such that the determination of the local regression parameters is a well posed problem. The regularisation of the model can be controlled in a distinct manner using only two user-defined parameters. Examples from an industrial problems illustrate the efficiency of the proposed algorithm.


  1. Babuska, R., Verbruggen, H.: An overview of fuzzy modeling for control. control Engineering Practice 4 (1996) 1593 - 1606
  2. Johansen, T.A., Foss, B.A.: Operating regime based process modeling and identification. Computers and Chemical Engineering 21 (1997) 159 - 176
  3. Babuska, R.: Recent Advances in Intelligent Paradigms and Applications. Springer-Verlag, Heidelberg (2002)
  4. Ren, X., Rad, A., Chan, P., Wai, L.: Identification and control of continuous-time nonlinear systems via dynamic neural networks. Industrial Electronics, IEEE Transactions on 50 (2003) 478-486
  5. Nelles, O.: Nonlinear System Identification. 1st edn. Springer Verlag (2002)
  6. Bezdek, J., Tsao, E.K., Pal, N.: Fuzzy kohonen clustering networks. In: IEEE International Conference on Fuzzy Systems, IEEE (1992) 1035 - 1043
  7. Yen, J., L., W.: Application of statistical information criteria for optimal fuzzy model construction. IEEE Transactions on Fuzzy Systems 6 (1998) 362 - 372
  8. Gath, I., Geva, A.: Unsupervised optimal fuzzy clustering. IEEE Transactions on Pattern Analysis and Machine Intelligence 11 (1989) 773-781
  9. Abonyi, J., Babuska, R., Szeifert, F.: Modified gath-geva fuzzy clustering for identification of takagi-sugeno fuzzy models. In: IEEE Tarnsactions on systems, man. and Cybernetics. Volume 32., IEEE (2002) 612-621
  10. Trabelsi, A., Lafont, F., Kamoun, M., G., E.: Identification of nonlinear multivariable systems by adaptive fuzzy takagi-sugeno model. International Journal of Computational Cognition (http://www.YangSky.com/yangijcc.htm) 2 (2004) 137-153
  11. Nelles, O., Isermann, R.: Basis function networks for interpolation of local linear models. In: IEEE Conference on Decision and Control (CDC). (1996) 470-475
  12. Bittermann, A., Kranawetter, E., J., K., B., L., T., E., Altenstrasser, H., Koegeler, H., Gschweitl, K.: Emissionsauslegung des dieselmotorischen fahrzeugantriebs mittels doe und simulationsrechnung. Motorentechnische Zeitschrift (2004)
  13. Hong, X., Sharkey, P., Warwick, K.: A robust nonlinear identification algorithm using press statistics and forward regression. In: IEEE Transactions on Neural Networks. Volume 14., IEEE (2003) 454 - 458
  14. Jakubek, S., Strasser, T.: Artificial neural networks for fault detection in large-scale data acquisition systems. Engineering Applications of Artificial Intelligence 17 (2004) 233-248

Paper Citation

in Harvard Style

Jakubek S. and Keuth N. (2005). A New Training Algorithm for Neuro-Fuzzy Networks . In Proceedings of the 1st International Workshop on Artificial Neural Networks and Intelligent Information Processing - Volume 1: ANNIIP, (ICINCO 2005) ISBN 972-8865-36-8, pages 23-34. DOI: 10.5220/0001180200230034

in Bibtex Style

author={Stefan Jakubek and Nikolaus Keuth},
title={A New Training Algorithm for Neuro-Fuzzy Networks},
booktitle={Proceedings of the 1st International Workshop on Artificial Neural Networks and Intelligent Information Processing - Volume 1: ANNIIP, (ICINCO 2005)},

in EndNote Style

JO - Proceedings of the 1st International Workshop on Artificial Neural Networks and Intelligent Information Processing - Volume 1: ANNIIP, (ICINCO 2005)
TI - A New Training Algorithm for Neuro-Fuzzy Networks
SN - 972-8865-36-8
AU - Jakubek S.
AU - Keuth N.
PY - 2005
SP - 23
EP - 34
DO - 10.5220/0001180200230034