Abstract
To avoid the need to pre-process noisy data, two special denoising layers based on wavelet multiresolution analysis have been integrated into layered neural networks. A gradient-based learning algorithm has been developed that uses the same cost function to set both the neural network weights and the free parameters of the denoising layers. The denoising layers, when integrated into feedforward and recurrent neural networks, were validated on three time series prediction problems: the logistic map, a rubber hardness time series, and annual average sunspot numbers. Use of the denoising layers improved the prediction accuracy in both cases.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
Masters T (1995) Neural, novel & hybrid algorithms for time series prediction. Wiley, Toronto
Weigend AS, Gershenfeld NA (1994) Time series prediction: forecasting the future and understanding the past. Addison Wesley, Reading, MA
Forsee FD, Hagan MT (1997) Gauss–Newton approximation to Bayesian regularization. In: Proc 1997 IEEE Int Joint Conf on Neural Networks, Houston, TX, 9–12 June 1997, pp 1930–1935
Drucker H, Cun YL (1992) Improving generalization performance using double backpropagation. IEEE T Neural Networ 3(6):991–997
Ster B (2003) Latched recurrent neural network. Electrotech Rev 70(1–2):46–51 (http://ev.fri.uni-lj.si/online.html)
Donoho DL (1995) De-noising by soft-thresholding. IEEE T Inform Theory 41:613–627
Prochazka A, Mudrova M, Storek M (1998) Wavelet use for noise rejection and signal modelling. In: Prochazka A et al (eds) Signal analysis and prediction. Birkhauser, Boston, MA, pp 215–226
Zhang Q, Benveniste A (1992) Wavelet networks. IEEE T Neural Networ 3:889–898
Bakshi BR, Stephanopoulos G (1993) Wave-net: a multiresolution hierarchical neural network with localized learning. AIChE J 39:57–81
Daubechies I (1992) Ten lectures on wavelets. SIAM, Philadelphia, PA
Lotric U, Dobnikar A (2003) Matrix formulation of multilayered perceptron with denoising unit. Electrotech Rev 70:221–226 (http://ev.fri.uni-lj.si/online.html)
Lotric U (2004) Wavelet based denoising integrated into multilayered perceptron. Neurocomp (in press)
Haykin S (1999) Neural networks: a comprehensive foundation, 2nd edn. Prentice-Hall, Englewood Cliffs, NJ
Press WH, Teukolsky SA, Vetterling WT, Flannery BP (1992) Numerical recipes in C, 2nd edn. Cambridge University Press, Cambridge, UK
Trentin E (2001) Networks with trainable amplitude of activation functions. Neural Networks 14:471–493
Williams RJ, Zipser D (1989) A learning algorithm for continually running fully recurrent neural networks. Neural Comput 1:270–280
Schuster HG (1984) Deterministic chaos: an introduction. Physik, Weinheim, Germany
Sunspot Index Data Center (2001) Yearly definitive sunspot number. Sunspot Index Data Center, Royal Observatory of Belgium, Brussels http://sidc.oma.be
Acknowledgements
The project has been funded in part by the Slovenian Ministry of Education, Science and Sport under Grant No. Z2–3040.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Lotric, U., Dobnikar, A. Predicting time series using neural networks with wavelet-based denoising layers. Neural Comput & Applic 14, 11–17 (2005). https://doi.org/10.1007/s00521-004-0434-z
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00521-004-0434-z