An Adaptive Neural Network in Wavelet Space for Time-Series Pediction.pdf

文档介绍：AN ADAPTIVE WORK IN WAVELET SPACE FOR TIME-SERIES PREDICTION
Fu-Chiang Tsui'I2 Ching-ChungLiz Mingui Sun' Robert .''2 '
Laboratory putational Neuroscience, Departments of Neurological Surgery'
and Electrical Engineering'
University of Pittsburgh, Pittsburgh, PA 15213-2582, USA,
tsui@ .pitt .edu Phone: [phone] FAX: -692-5921
ABSTRACT Appendix. The propositi'on states that the projection function at
the coarsest scale level is continuous in its coefficients; this assures
An adaptive recurrent work (ARNN) in wavelet coeffi-
that the smaller the error in the predicted coefficients is, the less
cient puted from the discrete wavelet transform (DWT)
the deviation in the predicted projection function will be.
is presented in this paper for generating an adaptive, long-term,
coarse resolution prediction of a time series. The weights inside 2. APPROACH
the ARNN are updated by the ing data, ., work mod-
ifies itself with time. With the aid of the newly developed DWT of The DWT of Cai-Wang[ putes wavelet coefficients from
Cai-Wang, this ARNN is efficient and takes less time to train than coarse to fine scales by applying the 4-th order €3-spline function
a NN in data space since it deals only with wavelet coefficients to interpolate data in the time domain. It allows a description
instead of raw data. Results are