Scalenet - multiscale neural network architecture for time series prediction

Amir B. Geva, Andres K. Altman

Research output: Contribution to conferencePaperpeer-review

Abstract

The effectiveness of multiscale neural network (NN) architecture for time series prediction of nonlinear dynamic systems has been investigated. The prediction task is simplified by decomposing the time series into separate scales of wavelets, and predicting each scale by a separate multilayer perceptron NN. The different scales of the wavelet transform provides an interpretation of the series structures and information about the history of the series, using fewer coefficients than other methods. In the next stage, the predictions of all the scales are combined, applying another perceptron NN, in order to predict the original time series. Each network is trained by the back-propagation algorithm using the Levenberg-Marquadt method. The weights and biases are initialized by new clustering methods, which improved the prediction results compared to random initialization. Three sets of data were analyzed: the sunspots' benchmark, fluctuations in a far-infrared laser and a numerically generated series (set A and D in the Santa Fe competition[2]). Taking the ultimate goal to be the accuracy of the prediction, we find that our suggested architecture outperforms traditional nonlinear statistical approaches.

Original languageEnglish
Pages243-246
Number of pages4
StatePublished - 1 Dec 1996
EventProceedings of the 1996 19th Convention of Electrical and Electronics Engineers in Israel - Jerusalem, Isr
Duration: 5 Nov 19966 Nov 1996

Conference

ConferenceProceedings of the 1996 19th Convention of Electrical and Electronics Engineers in Israel
CityJerusalem, Isr
Period5/11/966/11/96

ASJC Scopus subject areas

  • Engineering (all)

Fingerprint

Dive into the research topics of 'Scalenet - multiscale neural network architecture for time series prediction'. Together they form a unique fingerprint.

Cite this