title: Combining Neural Network Forecasts on Wavelet-Transformed Time Series
creator: Aussem, Alex
creator: Murtagh, Fionn
subject: Artificial Intelligence
subject: Dynamical Systems
subject: Machine Learning
subject: Neural Nets
subject: Speech
subject: Statistical Models
description: We discuss a simple strategy aimed at improving neural network prediction accuracy, based on the combination of predictions at varying resolution levels of the domain under investigation (here: time series). First, a wavelet transform is used to decompose the time series into varying scales of temporal resolution. The latter provide a sensible decomposition of the data so that the underlying temporal structures of the original time series become more tractable. Then, a Dynamical Recurrent Neural Network (DRNN) is trained on each resolution scale with the temporal-recurrent backpropagation (TRBP) algorithm. By virtue of its internal dynamic, this general class of dynamic connectionist network approximates the underlying law governing each resolution level by a system of nonlinear difference equations. The individual wavelet scale forecasts are afterwards recombined to form the current estimate. The predictive ability of this strategy is assessed with the sunspot series.
date: 1997
type: Journal (Paginated)
type: PeerReviewed
format: application/postscript
identifier: http://cogprints.org/551/2/ConnectScience.ps
identifier:   Aussem, Alex and Murtagh, Fionn  (1997) Combining Neural Network Forecasts on Wavelet-Transformed Time Series.  [Journal (Paginated)]     
relation: http://cogprints.org/551/